ReplicationControl SQL Option User s Manual

Transcription

1 NEC Storage Software ReplicationControl SQL Option User s Manual IS006-17E

2 NEC Corporation No part of the contents of this book may be reproduced or transmitted in any form without permission of NEC Corporation. The contents of this book may be modified without notice in the future.

3 Preface This manual describes the NEC Storage ReplicationControl SQL Option. By being applied in combination with the data replication function (NEC Storage DynamicDataReplication or NEC Storage RemoteDataReplication) or the snapshot function (NEC Storage DynamicSnapVolume), this product makes it possible to create a snapshot, without stopping operations, on the Microsoft SQL Server database constructed on the NEC Storage series disk array subsystem. This snapshot can be used for backup operations and parallel processing. This enables 24-hour uninterrupted operation of the Microsoft SQL Server database. If an error occurs in a master volume, base-volume, or the Microsoft SQL Server database, you can perform recovery with backups created using the data replication function or snapshot function and apply the transaction log backup to restore the Microsoft SQL Server database. This manual is intended for readers who already possess the basic knowledge about Microsoft SQL Server as well as experience with backup and recovery operations. The data replication function and the snapshot function for the NEC Storage disk array subsystems, which are used in this manual, are explained in the NEC Storage Software Data Replication User s Manual (Function Guide) (IS015) and the NEC Storage Software Snapshot User s Manual (Function Guide) (IS030), respectively. For information related to Microsoft SQL Server, refer to the Microsoft SQL Server Books Online. Refer to the NEC Storage Software Manual Guide (IS901) for the overview of NEC Storage and the related manuals. Remarks 1. This manual explains functions implemented by the following program products: NEC Storage Manager and NEC Storage BaseProduct NEC Storage ControlCommand (Note) NEC Storage ReplicationControl SQL Option NEC Storage DynamicDataReplication NEC Storage RemoteDataReplication NEC Storage DynamicSnapVolume Note: NEC Storage ControlCommand is a program product that has integrated the following five functions. ReplicationControl SnapControl ReplicationControl/DisasterRecovery ProtectControl PowerControl

4 2. This manual is applicable to the program products of the following versions: NEC Storage Manager Ver9.5 NEC Storage BaseProduct Ver9.5 NEC Storage ControlCommand Ver9.5 NEC Storage ReplicationControl SQL Option Ver Unless otherwise stated, the NEC Storage Manager is referred to as ism or Storage Manager in the text of this manual. Also, the NEC Storage series disk array subsystem is referred to as a disk array. 4. The following descriptions in the text of this manual refer to the corresponding products unless otherwise stated. 5. The NEC Storage series disk array subsystem is referred to as a disk array in this manual unless clearly specified. Also, the following terms refer to the corresponding NEC Storage hardware products. Term NEC Storage Hardware Product xxx series or xxxx series Sxxx or Sxxxx * xxx and xxxx represent the model number. NEC Storage xxx series or NEC Storage xxxx series NEC Storage Sxxx or NEC Storage Sxxxx 6. The following descriptions in the text of this manual refer to the corresponding manuals unless otherwise stated. User's Manual Messages Handbook Description ReplicationControl SQL Option ControlCommand DynamicDataReplication DynamicSnapVolume RemoteDataReplication Description Data Replication User's Manual (Function Guide) Data Replication User s Manual (Installation and Operation Guide for Windows) Snapshot User's Manual (Function Guide) Snapshot User s Manual (Installation and Operation Guide for Windows) ControlCommand Command Reference Corresponding Product NEC Storage ReplicationControl SQL Option NEC Storage ControlCommand NEC Storage DynamicDataReplication NEC Storage DynamicSnapVolume NEC Storage RemoteDataReplication Corresponding Manual NEC Storage Software NEC Storage Manager User's Manual (IS004) NEC Storage Software Messages Handbook (IS010) NEC Storage Software Data Replication User's Manual (Function Guide) (IS015) NEC Storage Software Data Replication User s Manual (Installation and Operation Guide for Windows) (IS016) NEC Storage Software Snapshot User's Manual (Function Guide) (IS030) NEC Storage Software Snapshot User s Manual (Installation and Operation Guide for Windows) (IS031) NEC Storage Software ControlCommand Command Reference (IS041) 7. This manual is based on the Microsoft SQL Server 2005 Enterprise Edition, Microsoft SQL Server 2005 Standard Edition, Microsoft SQL Server 2008 Enterprise, Microsoft SQL Server 2008 Standard, Microsoft SQL Server 2008 R2 Enterprise, Microsoft SQL Server 2008 R2 Standard, Microsoft SQL Server 2012 Enterprise, Microsoft SQL Server 2012 Standard, Microsoft SQL Server 2014 Enterprise, Microsoft SQL Server 2014 Standard, Microsoft SQL Server 2016 Enterprise Edition, and Microsoft SQL Server 2016

5 Standard Edition. 8. The terms data replication function and snapshot function used in this manual refer to the functions provided by DynamicDataReplication, RemoteDataReplication, DynamicSnapVolume, ReplicationControl, and SnapControl of the NEC Storage software series. They differ from the replication function provided by Microsoft SQL Server. 9. The term Microsoft SQL Server in this manual refers to any one of Microsoft SQL Server 2005 Enterprise Edition, Microsoft SQL Server 2005 Standard Edition, Microsoft SQL Server 2008 Enterprise, Microsoft SQL Server 2008 Standard, Microsoft SQL Server 2008 R2 Enterprise, Microsoft SQL Server 2008 R2 Standard, Microsoft SQL Server 2012 Enterprise, Microsoft SQL Server 2012 Standard, Microsoft SQL Server 2014 Enterprise, Microsoft SQL Server 2014 Standard, Microsoft SQL Server 2016 Enterprise, and Microsoft SQL Server 2016 Standard. 10. The term MSCS in this manual refers to the Microsoft Cluster Server. 11. The term MSFC in this manual refers to the Microsoft Failover Cluster. 12. Trademarks and registered trademarks Microsoft, Windows, and Windows Server are trademarks or registered trademarks of Microsoft Corporation in the United States and other countries. SQL Server is a registered trademark of Microsoft Corporation in the Unites States and other countries. VERITAS, Backup Exec, VERITAS Backup Exec, NetBackup, and VERITAS NetBackup are trademarks or registered trademarks of VERITAS Software Corporation in the United States and other countries. ARCserve is a registered trademark of Computer Associates International, Inc. in the United States. Other product names and company names, etc. are trademarks or registered trademarks of the associated companies. 13. In this document, the capacity is calculated based on units of 1024 (for example 1 KB = 1024 bytes) unless otherwise specified. 1st Edition in January th Edition in November 2016

6 Contents Chapter 1 Data Replication Creation Functions of Disk Arrays Overview of Data Replication Creation Functions Data Replication Function Snapshot Function... 5 Chapter 2 Overview of ReplicationControl SQL Option Product Function and Features System Configuration Examples of Use of Data Replication Function Backup Secondary Use for Search Operation and Test Operation Examples of Use of Snapshot Function Backup Secondary Use for Search Operation and Test Operation Example of Combined Use of Data Replication Function and Snapshot Function Backup Chapter 3 Database System Configuration Database Configuration Application to Backup System Backup Types Parallel Operations in Other Nodes Parallel Operation for Search Operation (Standby Server) Parallel Operation for Update Operation Application of Transaction Log Database Operated in Parallel Operation Automatization Chapter 4 Functions of ReplicationControl SQL Option Snapshot Backup Snapshot Backup Operations Snapshot Restore Operations Commands Snapshot Backup Command Snapshot Restore Command Metafile Handling Access Restriction Chapter 5 Installation and Setting of ReplicationControl SQL Option Setting of ReplicationControl SQL Option Logical Disk Preparation Database Environment Creation Disk Assignment for Meta Setting of Microsoft SQL Server Change of Maximum Number of Worker Threads Chapter 6 Operations of ReplicationControl SQL Option Data Replication Operations Tape Backup and Recovery Disk Backup and Recovery Environment Where System Databases Are Mixed Secondary Use for Search Operation Secondary Use for Update Operation Snapshot Operations Tape Backup and Recovery Disk Backup and Recovery Secondary Use for Search Operation i

7 Chapter 7 Measures for Errors Error Detection Method Errors Related to Database Configurations Errors Related to Metafile Configurations Collecting Information of Error of Unknown Cause Appendix A Notes Index ii

8 This page is intentionally left blank. iii

9 Chapter 1 Data Replication Creation Functions of Disk Arrays Chapter 1 Data Replication Creation Functions of Disk Arrays To centrally manage the enormous amount of information accumulated by business operations and promote its effective and efficient utilization, it is necessary to have a storage system that realizes a highly advanced processing feature, large capacity memory and high reliability. As a function which uses business data increasing day to day effectively and efficiently, improves the availability and operability of the corporate system, and supports 24-hour, 365-day system operation, a data replication creation function based on the storage system is essential. This chapter outlines the data replication creation functions provided by disk arrays. 1.1 Overview of Data Replication Creation Functions Data replication creation functions available with disk arrays are mainly divided into two types. You can use these data replication creation functions to construct and operate an effective system using the features of these functions. Data replication function The data replication function creates replications of a master volume defined in logical disk units in volume units. This function creates a replication volume as a volume physically independent of the relevant master volume. For this reason, the replication volume shows good performance and reliability when used and this function is suitable for secondary use and backup operations for which high performance and reliability are required. Snapshot function The snapshot function creates a replication (snapshot image) at a certain time of a base-volume defined in logical disk units in volume units. This function creates a snapshot-volume as a volume containing only the differential data from the base-volume that is managed by the disk array. For this reason, the snapshot-volume can be operated using a small capacity and this function is suitable for operations which place importance on the disk array capacity cost. 1

10 Chapter 1 Data Replication Creation Functions of Disk Arrays The following table lists differences between the data replication function and snapshot function. Table 1-1 Comparison of the Data Replication Function and Snapshot Function Replication function Snapshot function Required capacity of a replication Unit for referencing a replication Access to replications Whether the master volume or base-volume is affected by access to a replication Logical error Physical error Same capacity as that of the master volume Volume unit Read/write enabled Mounting on own/other servers enabled Not affected The master volume can be restored from the replication volume. The master volume can be restored from the replication volume. Only capacity of update data (differential data) of the base-volume Volume unit Read/write enabled Mounting on own/other servers enabled Affected The base-volume can be restored from snapshot-volumes. The base-volume cannot be restored from snapshot-volumes (snapshot-volume data is erased). 2

11 Chapter 1 Data Replication Creation Functions of Disk Arrays 1.2 Data Replication Function The data replication function, which is installed on a disk array, creates a replication volume (RV) for a master volume (MV). The MV and RV can be synchronized (replication) and separated (separation) at any given time. When the MV and RV are synchronized, their contents coincide with each other; any update to the MV is immediately reflected on the RV. After separated from the MV at any given time, the RV can be used as a replication of the MV. To instruct the data replication function to perform synchronization, separation, and other operations for the MV and RV, issue commands from the application server and use the ism client function. To create an RV using the data replication function, either of the following functions must be installed on the disk array: DynamicDataReplication (DDR) Creates an RV for the data replication function on the same disk array. RemoteDataReplication (RDR) Creates an RV for the data replication function between different disk arrays. Management Server Client Application Server Backup Server, etc. Master Operations Backup, Search, etc. Storage Manager SAN Operations of Data Replication Function RemoteDataReplication LAN Master Volume (MV) DynamicDataReplication Replication Volume (RV) Master Volume (MV) Replication Volume (RV) Disk Array Figure 1-1 Data Replication Function The following effects can be expected by using RVs created by the data replication function: Enabling uninterruptible system operation or substantially reducing the operation-stopped time when data is backed up Online data replications can be created in a short time. These replications can be used as backups as they are. Preventing the performance of access to the master database from degrading when data is backed up during system operation 3

12 Chapter 1 Data Replication Creation Functions of Disk Arrays Creating an RV does not affect the application server or operations because it is performed on the disk array. The RV can be used independently of the MV, so you do not need to worry about online response becoming slower during backup operation. Making the construction of a test environment easier and improving efficiency by parallel processing of data update operation and data reference operation An RV can be used to construct an evaluation environment easily using actual operation data. The RV can also be used for search operation and batch processing to perform parallel processing efficiently without considering interference or resource conflict with update operation on the MV. 4

13 Chapter 1 Data Replication Creation Functions of Disk Arrays 1.3 Snapshot Function The snapshot function, which is installed on a disk array, creates a replication of a base-volume (BV) on a snapshot-volume (SV) as a snapshot image. As replication data created by the snapshot function, only differential data from the source BV is stored and managed on the disk array. For this reason, replication data can be maintained using a small capacity and SVs can be used at low cost. You can instantly create a snapshot image at any desired time. You can also store created snapshot images on multiple SVs as time-series data for different generations. Link these SVs to a link-volume (LV) which can be accessed from servers to use them. To instruct the snapshot function to perform operations for the BV and LV such as the creation of snapshots and link connection, issue commands from the application server and use the ism client function. To create an SV using the snapshot function, the following function must be installed on the disk array: DynamicSnapVolume Creates an SV for the snapshot function on the disk array. Management Server Client Application Server Backup Server, etc. Master Operations Backup, Search, etc. Storage Manager SAN Operations of Snapshot Function LAN DynamicSnapVolume Base- Volume (BV) Snapshot- Volume (SV1) Snapshot- Volume (SV2) First generation (time t1) Second generation (time t2) LV Multiple generations Disk Array Snapshot- Volume (SVn) Nth generation (time tn) Figure 1-2 Snapshot Function 5

14 Chapter 1 Data Replication Creation Functions of Disk Arrays The following effects can be expected by using SVs created by the snapshot function: Enabling uninterruptible system operation or substantially reducing the operation-stopped time when data is backed up Online data replications can instantly be created. These replications can be used as backups as they are. Replications created at specific times on specific past days can be stored as data for multiple generations. Improving the efficiency in using the disk capacity Created replications, which contain only differential data, are stored and managed on the disk array. For this reason, the replication data can be used using a small capacity and at low cost. Making the construction of a test environment easier and improving efficiency by parallel processing of data reference operation Replication data can be used to construct an evaluation environment easily using actual operation data. The replication data can also be used as secondary data for search operation to perform parallel processing efficiently. 6

15 Chapter 2 Overview of ReplicationControl SQL Option Chapter 2 Overview of ReplicationControl SQL Option 2.1 Product Function and Features By applying ReplicationControl SQL Option product in combination with the data replication function (DynamicDataReplication or RemoteDataReplication) or snapshot function (DynamicSnapVolume), the Microsoft SQL Server database built on disk arrays is able to create a snapshot image of the database without suspending operations. Also, the snapshot is available in backup or parallel processing. These features enable round-the-clock, uninterruptible operation using the Microsoft SQL Server database. If an error occurs on the master volume, base-volume or Microsoft SQL Server database, snapshot backups created using the data replication function and snapshot function can be used to perform recovery and the transaction log backup can be applied to restore the Microsoft SQL Server database. 2.2 System Configuration When combined with the data replication function Application Server ReplicationControl SQL Option ControlCommand (ReplicationControl) Microsoft SQL Server Backup Server Backup Software ControlCommand (ReplicationControl) Client Replication Operation Management Server (FC-switch) Tape Drive Storage Manager LAN AccessControl BaseProduct Disk Array DB (MV) RemoteDataReplication DynamicDataReplication Replication DB (RV) When combined with the snapshot function 7

16 Chapter 2 Overview of ReplicationControl SQL Option Application Server ReplicationControl SQL Option ControlCommand (SnapControl) Microsoft SQL Server Backup Server Backup Software ControlCommand (SnapControl) Client Snapshot Operation Management Server (FC-switch) Tape Drive Storage Manager LAN AccessControl BaseProduct DB (BV) DynamicSnapVolume LV Snapshot DB (SV) Disk Array Figure 2-1 System Configuration ReplicationControl SQL Option operates in linkage with the following Microsoft software products to conduct snapshot operations: Microsoft SQL Server 2005 Enterprise Edition Microsoft SQL Server 2005 Standard Edition Microsoft SQL Server 2008 Enterprise Microsoft SQL Server 2008 Standard Microsoft SQL Server 2008 R2 Enterprise Microsoft SQL Server 2008 R2 Standard Microsoft SQL Server 2012 Enterprise Microsoft SQL Server 2012 Standard Microsoft SQL Server 2014 Enterprise Microsoft SQL Server 2014 Standard Microsoft SQL Server 2016 Enterprise Microsoft SQL Server 2016 Standard To install and use the data replication or snapshot, the following hardware devices are required. (1) Disk array To install the data replication function, a disk array on which the function of DynamicDataReplication or RemoteDataReplication is installed is required. To install the snapshot function, a disk array on which the function of DynamicSnapVolume is installed is required. (2) Management server and client The management server, on which ism is installed, monitors the disk array. This server manages the status of the disk array, and data replications and snapshots. 8

17 Chapter 2 Overview of ReplicationControl SQL Option (3) Application server and backup server The application server and backup server performs data replication and snapshot operations and backs up data in linkage with operations. The software performing data replication and snapshot consists of the following products. (1) ReplicationControl SQL Option Makes it possible to make a snapshot without suspending operation of the Microsoft SQL Server database. (2) ReplicationControl Provides the command that checks the data replication operation and state from the application server and backup server. (3) SnapControl Provides the commands that perform snapshot operations and check the status from the application server and backup server. (4) BaseProduct Consists of storage control software components which exercise basic control on the disk array and ism software components which enable the monitoring of the basic status of the disk array. (5) Storage Manager Provides a function for displaying the configuration and status of the disk array. Installing DynamicDataReplication and RemoteDataReplication makes it possible to use the replication management function incorporated in the ism. The replication management function displays the status of DynamicDataReplication and RemoteDataReplication on the disk array, and operates and sets data replications. Installing DynamicSnapVolume makes it possible to use the snapshot management function incorporated in ism. The snapshot management function displays the status of DynamicSnapVolume on the disk array, and operates and sets snapshots. (6) AccessControl Provides a function for setting disks which can be accessed from each of the application, backup, and other servers. This function prevents data from being destroyed due to unintentional access or being leaked or tampered by unauthorized access. (7) DynamicDataReplication Provides a function for creating replications of disks using the data replication function on the disk array. (8) RemoteDataReplication Provides a function for creating replications of disks using the data replication function between different disk arrays. (9) DynamicSnapVolume Provides a function for creating replications of disks using the snapshot function on the disk array. 9

18 Chapter 2 Overview of ReplicationControl SQL Option 2.3 Examples of Use of Data Replication Function Backup The operational application of making backup copies using a data replication function is described below. (1) Backup from an RV to a magnetic tape medium This is the backing up of an RV for an MV to a magnetic tape medium. By synchronizing (replicating) and separating the MV and the RV during the time period when the load on the business operations that use the Microsoft SQL Server is lower, it is possible to create an RV and back up copies onto magnetic tape media without suspending the operations. Since backup copies are made from RVs, MVs are not affected during backup. Application Server Backup Server Tape Drive Master Operations Backup Master Volume (MV) Replicate Separate Replication Volume (RV) Figure 2-2 Backup from RVs (1) During operation, keep the MV and the RV separate. (2) If the load on operations is lowered, replicate the data on the MV to the RV to synchronize them, and then separate them. (3) Execute the backup work concurrently with the business operation by using the separated RV. Also, after the backup is completed, the separated RV can be saved as a disk backup. 10

19 Chapter 2 Overview of ReplicationControl SQL Option (2) Using an RV as a disk backup This refers to the use of an RV as a backup copy of an MV. With this type of backup, the MV and RV are synchronized (replicated) and separated during the time period when the load on operations that use the Microsoft SQL Server is lower. This enables an RV to be created without suspending operations and a disk medium to be used as a backup. When restoring data from the backup data, it is possible to use backup data immediately after the start of restoration even if data from the RV to the MV is not completely copied. If data accessed from operations is not restored to the MV, restoration is performed while data is being copied from the RV. This makes it unnecessary for the user to consider whether to use the MV or RV. This makes it becomes possible to substantially reduce the data recovery time. Application Server Master Operations Replicate Master Volume (MV) Separate Replication Volume (RV1) Replication Volume (RV2) Replication Volume (RV3) Figure 2-3 Use of RV as Backup (1) During operation, keep all MVs and RVs (RV1, RV2, and RV3) separate. (2) If the load on operations is lowered, replicate the data on the MV to RV1 to synchronize them, and then separate them. (3) After that, make backup copies by using RV1, RV2 and RV3 in cycles. 11

20 Chapter 2 Overview of ReplicationControl SQL Option Secondary Use for Search Operation and Test Operation By creating a replication of the master database, the database replication can be used for search and test operations on another server. This allows parallel processing without these operations affecting each other. For search operation, the standby database function of Microsoft SQL Server is available. To use the standby database function, it is necessary to install ReplicationControl SQL Option on the server performing the search operation. Figure 2-4 Parallel Processing of Search Operation (1) Before starting search operation, replicate the data on the MV to the RV and separate them. (2) After that, applying the transaction log backup to the replication database can reflect the updates to the master database. Figure 2-5 Parallel Processing of Test Operation (1) During the daytime, keep the MV and RV separated. (2) At night, stop update operation, replicate the data on the MV to the RV to synchronize them, and then separate them. (3) The master and replication databases can be used in parallel. In this case, previous day s data is stored in the replication database. 12

21 Chapter 2 Overview of ReplicationControl SQL Option 2.4 Examples of Use of Snapshot Function Backup In this example, the snapshot function is used to create SVs of the BV for multiple generations and periodically back up copies onto magnetic tape media while using disk media as backups. By creating snapshots of the BV during the time period when the load on operations using Microsoft SQL Server is lower, it is possible to create SVs and save backups without suspending the operations. Application Server Backup Server Tape Drive Master Operations Backup Base- Volume (BV) Creating snapshots Snapshot- Volume (SV1) Snapshot- Volume (SV2) LV Linking the latest generation -1 generation (newest) -2 generation Snapshot- Volume (SV3) -3 generation Figure 2-6 Disk Backups Using SV and Backup to Magnetic Tape Media (1) Periodically (for example, once a day) create a snapshot (SV1) of the BV at night or during the time period when the load on operations is lower. Use SV1, SV2, and SV3 in cycles to create snapshots. (2) Link the newest created snapshot (SV1) to the LV to make a backup copy. Backup operation can be performed in parallel with the master operations. After the backup operation is completed, snapshots for multiple generations can be saved as disk backups. (3) If an operator or application error occurs and you want to restore logical data (restoration), you can use the disk backup for any desired generation to directly restore the BV from the snapshot (SV). This can substantially reduce the data restoration time. 13

22 Chapter 2 Overview of ReplicationControl SQL Option Secondary Use for Search Operation and Test Operation Replications of the master database can be used for search and test operations on another server. This allows parallel processing without these operations affecting each other. Data saved at specific times on specific past days can be saved for multiple generations to use it. For search operation, the standby database function of Microsoft SQL Server is available. To use the standby database function, it is necessary to install ReplicationControl SQL Option on the server performing the search operation. Application Server 1 Application Server 2 Application Server 3 Master Operations Test Operation Search Operation Base- Volume (BV) Snapshot- Volume (SV1) LV LV Creating snapshots Snapshot- Volume (SV2) Linking a desired generation Snapshot- Volume (SV3) Snapshot- Volume (SV4) Figure 2-7 Parallel Processing of Search Operation and Test Operation <Secondary use for search operation> (1) Create a snapshot (SV) of the BV periodically (for example, once every night). (2) Select the snapshot (SV) for the required generation and link it to the LV to use the SV. Applying the transaction log backup to the replication database can reflect the updates to the master database. <Secondary use for test operation> (1) At night, stop update operation and creates a snapshot (SV) of the BV. (2) During the daytime, linking the snapshot (SV) to the LV and using it on another server enables the use of the master and replication databases in parallel. In this case, previous day s data is stored in the replication database. 14

23 Chapter 2 Overview of ReplicationControl SQL Option 2.5 Example of Combined Use of Data Replication Function and Snapshot Function Backup In this example, the data replication function and snapshot function are concurrently used. A backup copy is saved on the RV once a day or week using the data replication function and the backup copy is also saved on tape media periodically to guard against disk and other physical errors. A backup copy is saved on each SV at intervals of several hours using the snapshot function to guard against logical errors such as file deletion due to an operator error. The following effects can be expected by concurrently using the data replication function and snapshot function: By creating an RV using the data replication function, if a physical error occurs, the MV can speedily be restored from the RV. The RV can be used to back up data onto tape media without affecting operations. By periodically creating an SV using the snapshot function to save the newest data, if a logical error occurs, the BV can be restored from the SV to minimize the damage. By saving data for multiple generations, data saved at the specific past time can be restored. Application Server Backup Server Tape Drive Master Operations Backup Creating snapshots Master Volume/ Base-Volume (MV/BV) Replicate Separate Snapshot- Volume (SV1) Replication Volume (RV) -1 generation (newest) Snapshot- Volume (SV2) -2 generation Snapshot- Volume (SV3) -3 generation Figure 2-8 Combined Use of Data Replication Function and Snapshot Function 15

24 Chapter 2 Overview of ReplicationControl SQL Option Notes on concurrent use When using the data replication function and snapshot function concurrently, note the following points: Data replication operations are affected by the snapshot state. When the BV (or MV) is restored from an SV, replication and restoration are disabled. Restoration using the data replication function is enabled only when all SVs paired with the BV (or MV) are inactive (snapshot inactive state). Table 2-1 Data Replication Operations and Snapshot States Snapshot state Snapshot Replication operation Active Inactive Deleting Restoring Replication - Separation Restoration - (*1) - - : Enabled (*1: Enabled only when all SVs are in the snapshot inactive state.) -: Disabled * Restoring includes two states: restore execution (rst/exec) and restore suspended. Snapshot operations are affected by the replication state. When the MV (or BV) is restored from an RV, snapshot creation and restoration are disabled. Snapshot restoration is enabled only when all RVs paired with the MV (or BV) are in the separated state. Table 2-2 Snapshot Operations and Data Replication States Replicate Separate Restore Execution Synchronized Execution Separated (*) Execution Synchronized Snapshot creation - - Snapshot restoration Snapshot deletion * This state includes separated, cancel, and fault. : Enabled -: Disabled Data replication state Snapshot operation 16

25 Chapter 3 Database System Configuration Chapter 3 Database System Configuration Database configuration and concept of backups for the Microsoft SQL Server are briefly described below. For details, refer to the Microsoft SQL Server Books Online. 3.1 Database Configuration The Microsoft SQL Server database consists of three ; the primary data file, the secondary data file and the log file. The primary data file and the secondary data file are also simply called data file. Database File1.mdf File2.ndf File3 File4 File1.ldf File2.ldf File group FILESTREAM Data file Log file Figure 3-1 Database Configuration Primary data file One database always has one primary data file. As the starting point of the database, the primary data file points to the locations of the secondary data and log (File1.mdf in the diagram above). The file name extension for the primary data file is.mdf. Secondary data file As a data file, the secondary data file sometimes does not exist and sometimes multiple secondary data exist (File2.ndf in the above figure). The extension for the secondary data file is normally.ndf. Log file Refers to the file where the log for recovery is stored. At least one file is required; sometimes, two or more also exist (File1.ldf and File2.ldf in the diagram above). The file name extension for the log file is normally.ldf. 17

26 Chapter 3 Database System Configuration File group On Microsoft SQL Server, this refers to a labeled aggregate of one or more. A unit of assignment or a unit of database management is called a file group. The default file group is PRIMARY. In Microsoft SQL Server 2008 or later, the FILESTREAM file group can be used to store FILESTREAM data (unstructured data such as text and image) as a data file. Although it is not necessary that file extensions, such as.mdf,.ndf, and.ldf, be used with the Microsoft SQL Server, those file extensions are useful for identifying file usage. By backing up the transaction log from the log file when a data error occurs, it is possible to recover data up to the point of error. This makes it necessary to allocate the data file and the log file in different logical volumes. In addition, to improve error resistibility, allocate the data file and the log file to the logical volumes assigned to different RANK or pool, respectively. For operational purposes, the ReplicationControl SQL Option processes all the, FILESTREAM data and file groups included in the target database. When specifying a pair of MV and RV, or of BV and SV with the parameters of snapshot backup and snapshot restore commands, be sure to specify them to all the included in the database. The data replication and snapshot functions operate in logical disk units. Therefore, when different database exist in a logical disk, all of the databases in the logical disk must be operated at the same time. Furthermore, the Microsoft SQL Server controls the connections of users for each instance. Therefore, it is not possible to refer to database information included in a different instance or information on the and configurations contained in the database. The ReplicationControl SQL Option checks that the specified databases properly correspond to the operation target logical disks. However, when database included in different instances exist in logical disks, it does not check the relationship between the databases and the logical disks. Therefore, when operating with multiple instances, be sure to specify all necessary databases and logical disks. The ReplicationControl SQL Option does not specify a static point on other than those of database. Do not allocate non-database to the backup target MV or BV. 18

27 Chapter 3 Database System Configuration 3.2 Application to Backup System The concept of using the snapshot backup created by ReplicationControl SQL Option as a backup to the master operations is described below along with the restoration model for Microsoft SQL Server Books Online Backup Types (1) Snapshot backup Refers to the type of backup used on the ReplicationControl SQL Option. Functionally the same as the conventional database s complete backup and file backup, etc., this type is capable of roll forwarding by using the transaction log backup. In the same manner as other backups, the snapshot backup and recovery are traced with MSDB. (2) Database complete backup Since the snapshot backup is used, a complete backup of the database is not made. (3) Database differential backup Since the snapshot backup is used, a differential backup of the database is basically not made. (4) Transaction log backup The database can be restored to the state immediately before the error occurred by combining the transaction log backup with the snapshot backup. In order to apply the transaction log after snapshot restoration, specify norecovery with the remod option. Perform the system database backup or restoration by using the normal backup method or with ReplicationControl or SnapControl by suspending the business operations of Microsoft SQL Server. For details on the timing of performing the system database backup, refer to the Microsoft SQL Server Books Online. 19

28 Chapter 3 Database System Configuration 3.3 Parallel Operations in Other Nodes An explanation on using the snapshot backup created by the ReplicationControl SQL Option for parallel operations of business on Microsoft SQL Server of other nodes is offered below Parallel Operation for Search Operation (Standby Server) By the snapshot backup, separated database can be used as a search database (standby database) by other servers. The search server must be used as a standby server. For details, refer to the Microsoft SQL Server Books Online. In this case, when performing snapshot restoration, specify norestore and then specify standby for the remod option. The database restored through snapshot restoration can be used as the database for the standby mode Parallel Operation for Update Operation The database once disconnected through snapshot backup can be used by other servers. In this case, by attaching the disconnected database, it can be used in the same manner as the ordinary database. It is not possible to apply the backup of the transaction log updated by the application server. To reflect the data updated by the application server, MV/RV synchronization and separation are required Application of Transaction Log To the database restored as the search database by executing snapshot restoration, the transaction log backup updated on the application server can be applied in the same way as to a database in the normal standby mode. By applying the transaction log, the search database data can be restored to the data at the time when the transaction log was copied as a backup. 20

29 Chapter 3 Database System Configuration Database Operated in Parallel When a database once disconnected through snapshot backup is operated in parallel through another node, the database cannot be used as a backup. To save it as a backup when necessary, either copy it onto a tape as a backup or use multiple RVs or SVs to backup in a different disk. 3.4 Operation Automatization The backup system, which used data replication or snapshot, requires a linkage operation between the business operation side and the backup server side. NetBackup is available the backup software provided with the linkage function between servers. To use ARCserve or BackupExec, job scheduling software is required to control job linkage operations between servers. 21

30 Chapter 4 Functions of ReplicationControl SQL Option Chapter 4 Functions of ReplicationControl SQL Option This chapter explains the operational outline, command operation methods, and metafile management of the snapshot backup that is a function of the ReplicationControl SQL Option operating on the application server. 4.1 Snapshot Backup Snapshot backup refers to a function that performs backup and restoration without suspending the database by using the functions provided by Microsoft SQL Server and data replication or snapshot function on the disk array. The snapshot backup minimizes the use of server s resource for backup purposes. The snapshot backup is conducted by the following functions: I/O freeze function Freezes the update I/O of Microsoft SQL Server to specify the static point of the database during the snapshot backup. Creation of metafile Refers to the file for storing Microsoft SQL Server s file configuration and transaction log state information, etc. The file is created during snapshot backup and the information is used for snapshot restoration. Data replication operation and snapshot operation Separates the MV and RV and restores data from the RV to the MV by using the functions of ReplicationControl when the data replication function is linked. Creates snapshots from the BV to SVs and restores data from SVs to the BV by using the functions of SnapControl when the snapshot function is linked. Moreover, if the update I/O for the backup target database is attempted during the snapshot backup, the Microsoft SQL Server freezes the update I/O for all databases included in the instance in which the backup target database is located. Ending the snapshot backup will automatically release the control of the update I/O for all databases and resume the update I/O function. 22

31 Chapter 4 Functions of ReplicationControl SQL Option Snapshot Backup Operations The snapshot backup operation includes: (1) freezing the update I/O of Microsoft SQL Server; (2) creating metafile; (3) flushing the MV or BV; (4) separating the MV and RV, or creation of snapshot from the BV to the SV; (5) defreezing the update I/O of Microsoft SQL Server. When combined with the data replication function Application Server (1) Freezing the update I/O for databases (2) Creating metafile (3) Flushing the MV (4) Separating the MV and RV (5) Defreezing the update I/O Backup Server Data file Separate Data file Log file Log file Master Volume (MV) Replication Volume (RV) Metafile When combined with the snapshot function Application Server Backup Server (1) Freezing the update I/O for databases (2) Creating metafile (3) Flushing the BV (4) Creating a snapshot from the BV to the SV (5) Defreezing the update I/O Data file Log file Creating snapshot Data file Log file Base-Volume (BV) Snapshot- Volume (SV) Link-Volume (LV) Metafile Figure 4-1 Operation to Execute Snapshot Backup 23

32 Chapter 4 Functions of ReplicationControl SQL Option Snapshot Restore Operations The snapshot restore operation includes: (1) acquiring the volume information of the MV or BV from the metafile; (2) unmounting the MV or BV; (3) restoring from the RV to the MV, or from the SV to the BV; (4) mounting the MV or BV; (5) restoring database. When combined with the data replication function Application Server Backup Server (1) Acquiring the business volume information from the metafile (2) Unmounting the MV (3) Restoring from the RV to the MV (4) Mounting the MV (5) Restoring database Data file Log file Restore Data file Log file Master Volume (MV) Replication Volume (RV) Metafile When combined with the snapshot function Application Server Backup Server (1) Acquiring the business volume information from the metafile (2) Unmounting the BV (3) Restoring from the SV to the BV (4) Mounting the BV (5) Restoring database Data file Log file Restore snapshot Data file Log file Base-Volume (BV) Snapshot- Volume (SV) Link-Volume (LV) Metafile Figure 4-2 Operation to Execute Snapshot Restoration 24

33 Chapter 4 Functions of ReplicationControl SQL Option 4.2 Commands Commands used on the ReplicationControl SQL Option are described below. Table 4-1 Lists of the ReplicationControl SQL Option Commands Command name Operation name Description ismsql_snapshotbkup Snapshot backup Backs up database. ismsql_snapshotrst Snapshot restore Restores database from backup. * Only users belonging to the Administrators group can perform the operations. In a Windows Server 2008 environment in which User Account Control (UAC) is enabled, to log on as a user other than the user belonging to the OS local system administrators group and perform command operations, elevate to the system administrator in advance. When using commands of ReplicationControl SQL Option, it is necessary to specify the setting of the I/O issue path to the disk array (IOPATH parameter) to directly issue the I/O commands to the disk array (DIRECT). For details on the operation settings, refer to the ControlCommand Command Reference Snapshot Backup Command Snapshot backup uses the ismsql_snapshotbkup command. By instructing Microsoft SQL Server, ismsql_snapshotbkup command freezes the update I/O, creates the metafile for the specified path and records the database configuration and the transaction log state. The command also separates the MV and RV where the specified database is stored or creates a snapshot from the BV to the SV, and creates a replication of the database. After this is completed, the command sends the completion instruction to Microsoft SQL Server, which in turn creates the history in the MSDB. Main Options The following options can be specified for the ismsql_snapshotbkup command regarding snapshot backup operation. (1) Specifying the database name (-db dbname or -dbfile dbfile_name) Specify the database name recognized by Microsoft SQL Server with a character string of 128 bytes or less. When simultaneously performing snapshot backups on multiple databases, specify multiple database names in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. 25

34 Chapter 4 Functions of ReplicationControl SQL Option (2) Specifying the metafile (-metafile metafile_name or -metafolder metafolder_name) A metafile name created during snapshot backup can be specified. When simultaneously performing snapshot backups on multiple databases, specify the name of the folder that stores meta. Under the specified folder name, multiple meta that correspond to each database on which a snapshot backup has been performed are created. The name of the meta is as follows: database-name_metafile.dat The metafile can be created for the shared disk, MV or BV, or local disk. However, it cannot be allocated to the logical disk containing the data and log of the database targeted for operation specified with -db or -dbfile. When a database is restored by executing snapshot restore, it is determined by referencing the contents of the metafile. For this reason, the metafile must be stored on a disk on which the operation target database is not stored. (3) Specifying the instance name (-server instance_name) The computer name and the instance name when connecting to a labeled instance can be specified. They can be specified with a character string of 272 bytes or less in computer-name\instance-name format. The length of a computer name must be 255 bytes or less, and that of an instance name is 16 bytes or less. When omitted, the connection is made to the default instance. The computer name must be either a local computer name, or a virtual computer name in the MSCS (Microsoft Cluster Server) or MSFC (Microsoft Failover Cluster) environment. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (4) Specifying the user ID/password The user ID and password to be connected to Microsoft SQL Server can be specified with a character string of 128 bytes or less. When omitted, the user name and password for the Windows login account is used. No entry in the password field is recognized as no password. Specifying from the command line (-uid userid -pass password): Specifies the user ID/password from the command line. Specifying from the keyboard (-input): When executing the command, interactively enter a user ID and password by using a keyboard. [User ID:] Enter a user ID. [Password:] Enter a password. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. 26

35 Chapter 4 Functions of ReplicationControl SQL Option (5) Specifying a database master key password (-dmkpass dmkpassword) When the database is encrypted, and the database master key is also encrypted with a password, specify that password with a character string of 128 bytes or less. When omitted, the system operates assuming that the database master key is not encrypted with any password or that the database is not encrypted. These specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (6) Canceling the consistency check (-nocheck) Specify this option to operate only part of the databases allocated to the MV or BV. When this option is specified, the consistency check is not performed. The consistency check refers to the verification whether all database on the MV or BV are specified as operation target. When the data replication function is used, this command with this option specified freezes the database without checking the synchronized state of the MV and RV. Before executing this command, complete synchronization to make sure the replicated state is set. (7) Specifying an operation target pair To use the data replication function, specify a MV and RV as an operation target pair. Specifying the MV (-mv volume -mvflg mv_flg): Specifies the MV volume and volume type. Specifying the RV (-rv volume -rvflg rv_flg): Specifies the RV volume and volume type. To use the snapshot function, specify a BV and SV as an operation target pair. Specifying the BV (-bv volume -bvflg bv_flg): Specifies the BV volume and volume type. Specifying the SV (-sv volume -svflg sv_flg): Specifies the SV volume and volume type if you want to specify a specific SV. Specifying the cyclic use of SVs (-cyclic): Specify this option to automatically select the SV for the oldest generation without specifying an SV for the cyclic use of SVs. For details on the specifiable volumes and volume type, refer to the ControlCommand Command Reference. If you want to operate multiple pairs at a time, you can write the pairs in a text file and specify the file. However, you cannot specify pairs for the data replication function and for the snapshot function in a file as operation target pairs. Specifying pairs at a time (-file file_name): You can use the replication operation file for the data replication function or the operation volume setting file for the snapshot function to specify multiple operation target pairs at a time. For details on the replication operation file and operation volume setting file, refer to the ControlCommand Command Reference. 27

36 Chapter 4 Functions of ReplicationControl SQL Option (8) Specifying RV access restriction after snapshot backup (-rvacc) When using the data replication function, access restriction on the RV after snapshot backup can be specified. Read Only (-rvacc ro): RV is possible only for reference purposes. Read/Write (-rvacc rw): RV is possible for reference or update purpose. Unless well versed in OS and file system operations for Windows, Read Only (-rvacc ro) should not be specified for the RV access restriction. For details, refer to 3.6 RV Access Restriction Setting in the Data Replication User s Manual (Installation and Operation Guide for Windows). (9) Specifying waiting for separate completion (-wait/-nowait) When using the data replication function, the program can be made to wait for the end of the command after executing the separation. Wait specified (-wait [second]): When the value is specified to second, the disk array s monitoring time interval becomes the specified time. The specifiable value is 1 to 30 seconds (integer number). When the second specification is omitted, the value specified for the RPL_WATCHDEV parameter in the replication operation option setting file is used. No wait (-nowait): After separate is executed, it is not waited for completion. To check the completion of the separation, use the ismrc_query or ismrc_wait command. (10) Specifying the time when the RV becomes available (-rvuse) When using the data replication function, it is possible to specify the time at which the RV becomes available after separation has been performed. This option is effective when the -nocheck option is specified and snapshot backup is executed without checking the synchronization state of the MV and RV. Normal separation (-rvuse complete): Makes the RV available after separate has been executed, the differential reflection of the MV and RV and the separation has been completed (Separate (completion)). Immediate separation (-rvuse immediate): Makes the RV available immediately after the start of separate while the differential reflection of the MV and RV is being performed after separate has been executed (Separate (immediate)). 28

37 Chapter 4 Functions of ReplicationControl SQL Option Using the default setting (-rvuse default): Uses the RVUSE parameter setting stored in the replication operation option setting file. The Separate (immediate) function can be used only when DynamicDataReplication Ver2 has been installed in the disk array. For details on this function, refer to the Data Replication User s Manual (Function Guide). Display information The ismsql_snapshotbkup command outputs the following messages: When using the data replication function [Start message] SnapshotBackup Start YYYY/MM/DD hh:mm:ss DataBase db_name MV:disk_number ld_name type volume_name path RV:disk_number ld_name type volume_name path ismsql_snapshotbkup: Info: ism13238: Freeze of database is started. YYYY/MM/DD hh:mm:ss Flush Start YYYY/MM/DD hh:mm:ss Flush Normal End YYYY/MM/DD hh:mm:ss Separate Start YYYY/MM/DD hh:mm:ss ismsql_snapshotbkup: Info: ism13239: All of database were thawed. YYYY/MM/DD hh:mm:ss [Separate in progress message] Separating... [Separate completion message] Separate Normal End YYYY/MM/DD hh:mm:ss 29

38 Chapter 4 Functions of ReplicationControl SQL Option [End message] SnapshotBackup Normal End DataBase db_name MV:disk_number ld_name volume_name path RV:disk_number ld_name volume_name path YYYY/MM/DD hh:mm:ss type type Explanation The separate in progress and separate completion messages are output only when waiting for separate completion ( -wait option) is specified. The output of the separate in progress message can be suppressed by setting the RPL_WAITMSG parameter in the replication operation option setting file. For details on the replication operation option setting file, refer to the ControlCommand Command Reference. SnapshotBackup Start YYYY/MM/DD hh:mm:ss Snapshot backup start message and start date and time SnapshotBackup Normal End YYYY/MM/DD hh:mm:ss Snapshot backup end message and end date and time DataBase db_name Database name MV:disk_number ld_name type volume_name path Physical disk number, logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the MV RV:disk_number ld_name type volume_name path Physical disk number, logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the RV ism13238: Freeze of database is started. YYYY/MM/DD hh:mm:ss Database freeze start message and start date and time ism13239: All of database were thawed. YYYY/MM/DD hh:mm:ss Database defreeze message and defreeze date and time Flush Start YYYY/MM/DD hh:mm:ss Flushing start message and start date and time 30

39 Chapter 4 Functions of ReplicationControl SQL Option Flush Normal End YYYY/MM/DD hh:mm:ss Flushing end message and end date and time Separate Start YYYY/MM/DD hh:mm:ss Separate start message and start date and time Separating... Separate in progress message The. display is dynamically updated until separation is completed. Separate Normal End YYYY/MM/DD hh:mm:ss Separate completion message and end date and time When using the snapshot function [Start message] SnapshotBackup Start YYYY/MM/DD hh:mm:ss DataBase db_name BV:ld_name type volume_name path SV:ld_name type (generation) attribute Delete Start YYYY/MM/DD hh:mm:ss ismsql_snapshotbkup: Info: ism13250: Snapshot has been deleted. SV:ld_name Delete Normal End YYYY/MM/DD hh:mm:ss ismsql_snapshotbkup: Info: ism13238: Freeze of database is started. YYYY/MM/DD hh:mm:ss Flush Start YYYY/MM/DD hh:mm:ss Flush Normal End YYYY/MM/DD hh:mm:ss Create Start YYYY/MM/DD hh:mm:ss Create Normal End YYYY/MM/DD hh:mm:ss ismsql_snapshotbkup: Info: ism13239: All of database were thawed. YYYY/MM/DD hh:mm:ss SnapshotBackup Normal End YYYY/MM/DD hh:mm:ss DataBase db_name BV:ld_name type volume_name path SV:ld_name type (generation) 31

40 Chapter 4 Functions of ReplicationControl SQL Option Explanation The relative generation number of the SV displayed at the start of snapshot backup indicates the generation number before a snapshot is created. An asterisk (*) is displayed for the generation attribute when the SV has a derived attribute. The relative generation number of the SV displayed at the end of the snapshot backup indicates the generation number after the snapshot is created and is always -1, which indicates the latest generation. The generation attribute is not displayed since the snapshot creation does not make the SV have a derived attribute. SnapshotBackup Start YYYY/MM/DD hh:mm:ss Snapshot backup start message and start date and time SnapshotBackup Normal End YYYY/MM/DD hh:mm:ss Snapshot backup end message and end date and time DataBase db_name Database name BV:ld_name type volume_name path Logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the BV SV:ld_name type (generation) attribute Logical disk name, OS type, relative generation number, and generation attribute of the SV ism13250: Snapshot has been deleted. SV:ld_name Snapshot delete message and logical disk name of the target SV ism13238: Freeze of database is started. YYYY/MM/DD hh:mm:ss Database freeze start message and start date and time ism13239: All of database were thawed. YYYY/MM/DD hh:mm:ss Database defreeze message and defreeze date and time Delete Start YYYY/MM/DD hh:mm:ss Snapshot delete start message and start date and time Delete Normal End YYYY/MM/DD hh:mm:ss Snapshot delete end message and end date and time Flush Start YYYY/MM/DD hh:mm:ss Flushing start message and start date and time Flush Normal End YYYY/MM/DD hh:mm:ss Flushing end message and end date and time Create Start YYYY/MM/DD hh:mm:ss Snapshot create start message and start date and time Create Normal End YYYY/MM/DD hh:mm:ss Snapshot create end message and end date and time 32

41 Chapter 4 Functions of ReplicationControl SQL Option Execution Conditions When performing snapshot backup, it is necessary to meet the following conditions. When using the data replication function (1) MVs and RVs targeted for operation is set as a pair. (2) The MV targeted for operation is registered in the volume list. (3) Snapshot backup is not executed when the replication function on the disk array on which the MV and RV targeted for operation are stored is in the freeze state. (4) Snapshot backup is executed only when the activity state of the pair targeted for operation is restore synchronous or replicate. (5) Snapshot backup is not executed when the specified MV is used as a base volume (BV) of the snapshot function, and it is in the restore state or restore fault state of the snapshot function. (6) All the data file and the log file included in the database exist in the MV targeted for operation. (7) The specified database name is unique. (8) The metafile-specified location is not the same disk as where the data file and log file of the database are stored. (9) The disk where the metafile is created has enough unused space to store the metafile. (10) Snapshot backup is not executed when a volume list is being created or updated. (11) A dynamic disk cannot be used. (12) No operation is performed for a pair registered in an AT-group. (13) Snapshot backup is not executed during operation in linkage with ism. When using the snapshot function (1) BV and SVs targeted for operation is set as a pair. (2) The BV targeted for operation is registered in the volume list. (3) Snapshot backup is not executed when the replication function on the disk array on which the BV and SVs targeted for operation are stored is in the freeze state. (4) Snapshot backup is not executed when the snapshot state of at least one SV paired with the BV targeted for operation is restore execution. (5) Snapshot backup is not executed when the BV targeted for operation is in the restore state of the data replication function. (6) Snapshot backup is not executed when the BV targeted for operation is in the snapshot fault state or restore fault state. (7) Snapshot backup cannot be executed when the SV targeted for operation is linked with the LV. (8) Snapshot backup cannot be executed when the guard specification is made for the SV targeted for operation. (9) Snapshot backup can be executed only when the SV targeted for operation is in the Not Ready state. (10) All the data and log included in the database exists in the BV targeted for operation. (11) The specified database name should be unique. (12) The metafile-storage location is not the same disk as where the data and log of the database targeted for operation are stored. (13) The disk where the metafile is to be created has enough unused space to store the metafile. (14) Snapshot backup is not executed when a volume list is being created or updated. 33

42 Chapter 4 Functions of ReplicationControl SQL Option (15) A dynamic disk cannot be used. (16) Snapshot backup is not executed during operation in linkage with ism Snapshot Restore Command Snapshot restore uses the ismsql_snapshotrst command. By executing ismsql_snapshotrst command, the restoration target master or base-volume is identified from the specified metafile to be restored from the RV to the MV, or from the SV to the BV. After restoration is executed, the database must be restored by giving the database restoration instruction to Microsoft SQL Server. Main Options With the ismsql_snapshotrst command, the following options can be specified with regard to the snapshot backup operation. (1) Specifying the database name (-db dbname or -dbfile dbfile_name) A database name recognized by Microsoft SQL Server is specified with a character string of 128 bytes or less. When simultaneously performing snapshot backups on multiple databases, specify multiple database names in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (2) Specifying the metafile (-metafile metafilename or -metafolder metafolder_name) The metafile name specified during snapshot backup is specified. When simultaneously performing snapshot restorations on multiple databases, specify the name of the folder that stores meta. However, you cannot specify a metafile allocated to the logical disk containing the data and log of the database targeted for operation specified with -db or -dbfile. (3) Specifying the instance name (-server instance_name) The computer name and the instance name when connecting to a labeled instance can be specified. They can be specified with a character string of 272 bytes or less in computer-name\instance-name format. The length of a computer name must be 255 bytes or less, and that of an instance name is 16 bytes or less. When omitted, the connection is made to the default instance. The computer name must be either a local computer name, or a virtual computer name in the MSCS (Microsoft Cluster Server) or MSFC (Microsoft Failover Cluster) environment. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. 34

43 Chapter 4 Functions of ReplicationControl SQL Option (4) Specifying the user ID/password The user ID and password to be connected to Microsoft SQL Server can be specified with a character string of 128 bytes or less. When omitted, the user name and password for the Windows login account is used. No entry in the password field is recognized as no password. Specifying from the command line (-uid userid -pass password): Specifies the user ID/password from the command line. Specifying from the keyboard (-input): When executing the command, interactively enter a user ID and password by using a keyboard. [User ID:] Enter a user ID. [Password:] Enter a password. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (5) Specifying a database master key password (-dmkpass dmkpassword) When the database is encrypted, and the database master key is also encrypted with a password, specify that password with a character string of 128 bytes or less. When omitted, the system operates assuming that the database master key is not encrypted with any password or that the database is not encrypted. These specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (6) Specifying restoration mode (-remod) The mode of database restoration can be specified. Recovering the database immediately (-remod recovery) Recovered as the database when snapshot backup was performed. Recovering the database in the state in which the transaction log can be applied. (-remod norecovery) Recovered as the database in the state in which the transaction log can be applied. The database cannot be used until transaction log is applied. Recovering the database as the database for search (-remod standby -refile refilename) Recovered as the database for search (standby database). The undo file (standby file) must be specified as well. The absolute path for the undo file (standby file) must be specified with a character string of 256 bytes or less. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. 35

44 Chapter 4 Functions of ReplicationControl SQL Option (7) Specifying that restoration is not executed (-norestore) Specify this option to restore only the database without restoring data from the RV to MV or from the SV to BV. You can use this option if you want to restore a test or search database. (8) Specifying database file transfer (-move os_file_name1 [, os_file_name2,...]) When the database is restored only as the test or search database, this option is used when the snapshot backup environment and the database file location differ. Specify a path name of the file, which makes up a database to be restored, with a character string of 512 bytes or less. Specify absolute path names of by separating with a comma. Those specifications can be described in the database operation file. For details on the database operation file, refer to the ControlCommand Command Reference. (9) Canceling the consistency check (-nocheck) Specify this option to operate only part of the databases allocated to the MV or BV. When this option is specified, the consistency check is not performed. The consistency check refers to the verification whether all database on the MV or BV are specified as operation target. (10) Specifying an operation target pair To use the data replication function, specify an MV and RV as an operation target pair. Specifying the MV (-mv volume -mvflg mv_flg): Specifies the MV s volume and volume type. Specifying the RV (-rv volume -rvflg rv_flg): Specifies the RV s volume and volume type. To use the snapshot function, specify a BV and SV as an operation target pair. Specifying the BV (-bv volume -bvflg bv_flg): Specifies BV s volume and volume type. Specifying the SV (-sv volume -svflg sv_flg): Specifies SV s volume and volume type. For details on the specifiable volumes and volume types, refer to the ControlCommand Command Reference. If you want to operate multiple pairs at a time, you can write the pairs in a text file and specify the file. However, you cannot specify pairs for the data replication function and for the snapshot function in a file as operation target pairs. Specifying pairs at a time (-file file_name): You can use the replication operation file for the data replication function or the operation volume setting file for the snapshot function to specify multiple operation target pairs at a time. For details on the replication operation file and operation volume setting file, refer to the ControlCommand Command Reference. 36

45 Chapter 4 Functions of ReplicationControl SQL Option (11) Specifying the copying range (-cprange) When using the data replication function, you can specify the copying range for restoration. Differential copy (-cprange diff): Copies only the differential area between the MV and RV. Full copy (-cprange full): Copies the full area of the RV. (12) Specifying the RV operation mode (-mode) When using the data replication function, you can specify the RV operation mode during the execution of restoration. RV update restoration (-mode update) Restores the update that has been performed on the MV while reflecting the update on the RV. After the restoration has been completed, the mode switches to the restoration complete state. RV protection restoration (-mode protect) Restores the update that has been performed on the MV without reflecting the update on the RV. After the restoration has been completed, the RV is automatically separated and the mode switches to the separation complete state. Using the default setting (-mode default) Uses the setting of the MODE parameter stored in the replication operation option setting file. The RV protection restoration function can be used only when either the DynamicDataReplication Ver2 or the RemoteDataReplication Ver2 is installed in the disk array. For details on this function, refer to the Data Replication User s Manual (Function Guide). (13) Specifying the Copy Control State (-cpmode) When using the data replication function, you can specify the Copy Control State for reflecting the update to the MV to the RV in the restore execution or restore synchronous state. This specification is effective for RV update restoration and is ignored for RV protection restoration. Synchronous copy (-cpmode sync): Copies data in the synchronous copy mode. Semi-synchronous copy (-cpmode semi): Copies data in the semi synchronous copy mode. This copy mode can be specified only for RemoteDataReplication pairs. Background copy (-cpmode bg): Copies data in the background copy mode. 37

46 Chapter 4 Functions of ReplicationControl SQL Option (14) Specifying the RV access restriction (-rvacc) When using the data replication function, you can specify the RV access restriction in the restore execution or restore synchronous state. When RV protection restoration is specified, the RV is separated and enters the Read/Write state automatically at the completion of restoration. Not Ready (-rvacc nr): The RV cannot be referenced or updated. Read Only (-rvacc ro): The RV can be referenced only. Unless well versed in OS and file system operations for Windows, specify Not Ready (-rvacc nr) for the RV access restriction. For details, refer to 3.6 RV Access Restriction Setting in the Data Replication User s Manual (Installation and Operation Guide for Windows). (15) Specifying waiting for restore completion (-wait/-nowait) When using the data replication function, you can specify the following option to make the command wait for the completion of restoration. Wait specified (-wait [second]): When the value is specified to second, the disk array s monitoring time interval becomes the specified time. The specifiable value is 1 to 30 seconds (integer number). When the second specification is omitted, the value specified for the RPL_WATCHDEV parameter in the replication operation option setting file is used. No wait (-nowait): Starts restoration only and does not wait for the completion of the restoration. To check the completion of the restoration, use the ismrc_query or ismrc_wait command. When using the snapshot function, you can specify the following option to make the command wait for the completion of restoration. Wait specified (-wait): Starts restoration and waits for the completion of the restoration. When using the data replication function, you can specify disk array s monitoring time interval in seconds. When using the snapshot function, however, you cannot specify disk array s monitoring time interval. A value in seconds is ignored if specified. No wait (-nowait): Starts restoration only and does not wait for the completion of the restoration. To check the completion of the restoration, use the ismsc_query command. 38

47 Chapter 4 Functions of ReplicationControl SQL Option (16) Specifying keeping the derived generation (-derivsv) When using the snapshot function, you can specify the keep mode of the derived generation snapshot data. Derived generation keep (-derivsv keep): Keeps the snapshot data of the derived generation SV. Derived generation delete (-derivsv delete): Deletes the snapshot data of the derived generation SV automatically. Default (-derivsv default): Follows the SNAP_RST_DERIVED parameter setting described in the snapshot function option setting file. When this argument is omitted, the operation is the same as when specifying default. The derived generation keep function is available when DynamicSnapVolume Ver3 is installed in the disk array. For details on this function, refer to the Snapshot User s Manual (Function Guide). Display information The ismsql_snapshotrst command outputs the following messages: When using the data replication function [Start message] SnapshotRestore Start YYYY/MM/DD hh:mm:ss DataBase db_name MV:disk_number ld_name type volume_name path RV:disk_number ld_name type volume_name path Restore Start YYYY/MM/DD hh:mm:ss [Restore in progress message] Restoring... [Restore completion message] Restore Normal End YYYY/MM/DD hh:mm:ss [End message] SnapshotRestore Normal End YYYY/MM/DD hh:mm:ss DataBase db_name MV:disk_number ld_name type volume_name path RV:disk_number ld_name type volume_name path 39

48 Chapter 4 Functions of ReplicationControl SQL Option Explanation The restore in progress and restore completion messages are output only when waiting for restore completion is specified. The output of the restore in progress message can be suppressed by setting the RPL_WAITMSG parameter in the replication operation option setting file. For details on the replication operation option setting file, refer to the ControlCommand Command Reference. SnapshotRestore Start YYYY/MM/DD hh:mm:ss Snapshot restore start message and start date and time SnapshotRestore Normal End YYYY/MM/DD hh:mm:ss Snapshot restore end message and end date and time DataBase db_name Database name MV:disk_number ld_name type volume_name path Physical disk number, logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the MV RV:disk_number ld_name type volume_name path Physical disk number, logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the RV Restore Start YYYY/MM/DD hh:mm:ss Restore start message and start date and time Restoring... Restore in progress message The. display is dynamically updated until restoration is completed. Restore Normal End YYYY/MM/DD hh:mm:ss Restore completion message and end date and time When using the snapshot function [Start message] SnapshotRestore Start YYYY/MM/DD hh:mm:ss DataBase db_name BV:ld_name type volume_name path SV:ld_name type (generation) attribute Restore Start YYYY/MM/DD hh:mm:ss [Restore completion message] Restore Normal End YYYY/MM/DD hh:mm:ss [End message] SnapshotRestore Normal End DataBase db_name BV:ld_name YYYY/MM/DD hh:mm:ss type 40

49 Chapter 4 Functions of ReplicationControl SQL Option volume_name path SV:ld_name type (generation) attribute Explanation The restore completion message is output only when waiting for restore completion is specified. The relative generation number of the SV displayed at the start of snapshot restoration indicates the generation number before a snapshot is restored. The relative generation number of the SV displayed at the end of the snapshot restoration indicates the generation number after the snapshot is restored. And it always indicates -1, which shows the latest generation. An asterisk (*) is displayed for the generation attribute when the SV has a derived attribute. SnapshotRestore Start YYYY/MM/DD hh:mm:ss Snapshot restore start message and start date and time SnapshotRestore Normal End YYYY/MM/DD hh:mm:ss Snapshot restore end message and end date and time DataBase db_name Database name BV:ld_name type volume_name path Logical disk name, OS type, mount point volume name, and path information (drive letter or NTFS folder path name) of the BV SV:ld_name type (generation) attribute Logical disk name, OS type, relative generation number, and generation attribute of the SV Restore Start YYYY/MM/DD hh:mm:ss Restore start message and start date and time Restore Normal End YYYY/MM/DD hh:mm:ss Restore completion message and end date and time Execution Conditions When performing snapshot restoration, it is necessary to meet the following conditions. When using the data replication function (1) MVs and RVs targeted for operation is set as a pair. (2) The operation target MV is registered in the volume correspondence table. (3) Snapshot restore is not executed when the replication function on the disk array on which the MV and RV targeted for operation are stored is in the freeze state. (4) Snapshot restore is executed only when the activity state of the pair targeted for operation is separated (separated). (5) When the MV targeted for operation is used as a BV, all SVs paired with the BV is in the snapshot 41

50 Chapter 4 Functions of ReplicationControl SQL Option inactive state. (6) Snapshot restore is not executed when the specified MV is used as a base volume (BV) of the snapshot function, and it is in the restore state or restore fault state of the snapshot function. (7) With regard to the pair set in the same disk array, the semi-synchronous copy mode is not specified. When specifying the semi-synchronous copy mode, it is for the pair set between disk arrays. (8) When performing restoration, all the data file and the log file included in the database exists in the RV targeted for operation. (9) The specified database name is unique. (10) The metafile-storage location should not be the same disk as where the data and log of the database targeted for restoration are stored. (11) Snapshot restore cannot be executed when a volume list is being created or updated. (12) A dynamic disk cannot be used. (13) The MV and RV targeted for operation should have been unmounted. (14) No operation can be performed for a pair registered in an AT-group. (15) Snapshot restore cannot be executed during operation in linkage with ism. When using the snapshot function (1) BVs and SVs targeted for operation should be set as a pair. (2) The BV targeted for operation should be registered in the volume list. (3) Snapshot restore cannot be executed when the replication function on the disk array on which the BV and SVs targeted for operation are stored is in the freeze state. (4) The BV and SVs targeted for operation should be in the snapshot active state. (5) Snapshot restore is not executed when the BV targeted for operation is in the snapshot fault state or restore fault state. (6) Snapshot restore cannot be executed with the setting in which the derived generation snapshot data is automatically deleted when there is an SV whose generation is later than that of the SV targeted for operation and the guard specification is made for that SV. (7) When snapshot restore is executed with the setting in which the derived generation snapshot data is automatically deleted, snapshot data whose generation is later than that of the SV targeted for operation will be automatically deleted. (8) Snapshot restore cannot be executed when the BV targeted for operation is used as an MV and is in the replicate or restore state of the data replication function. (9) Snapshot restore cannot be executed when the BV or SV targeted for operation is linked with an LV. (10) Snapshot restore can be executed only when the SV targeted for operation is in the Not Ready state. (11) For restoration, all the data and log included in the database should exist in the SV targeted for operation. (12) The specified database name should be unique. (13) The metafile-storage location should not be the same disk as where the data and log of the database are stored. (14) Snapshot restore cannot be executed when a volume list is being created or updated. (15) A dynamic disk cannot be used. 42

51 Chapter 4 Functions of ReplicationControl SQL Option (16) The BV targeted for operation should have been unmounted. (17) Snapshot restore cannot be executed during operation in linkage with ism. 4.3 Metafile Handling The metafile created when executing the snapshot backup is described below. One metafile is required for each backup. The metafile contains the operation target database s file configuration and the transaction log state information. The file size of a metafile is approximately several hundred KB. Its maximum size is approximately 1 MB. The metafile can be created on the shared disk, MV, BV or local disk. This metafile is necessary when recovering using the snapshot backup or operating the standby database on another node. Normally, the metafile is created on the shared disk because access to the metafile is easy. When the metafile is lost, snapshot restoration cannot be executed. Therefore, creating or saving it onto a highly reliable disk is recommended. There respective uses are described below. Shared disk A disk on the application or backup server is used as a shared disk and a metafile is created on the shared disk. The metafile created on the shared disk from the application server is backed up by accessing it from the backup server. This mode has the advantage of allowing both the application and backup servers to easily access the metafile via the shared disk. MV or BV When a shared disk is unavailable for network environment reasons, a metafile is created on the MV or BV and backed up using the data replication function or snapshot function. The metafile should be allocated to a disk other than the disk to which the database is allocated, however. On the MV or BV to which the database targeted for operation is allocated, the metafile cannot be created. After executing snapshot backup, back up the metafile. When using the data replication function, unmount the MV on which the metafile is created, separate the relevant pair, and create a replication of the metafile on the RV. When using the snapshot function, unmount the BV on which the metafile is created, create a snapshot, and create a replication of the metafile on the SV. To restore a database by executing snapshot restore, information in the metafile is required. For this reason, before executing snapshot restore, restore the metafile. Local disk To directly back up a metafile to a tape or disk on the application server, the metafile is created on a local disk of the application server. When using the snapshot backup as a disk backup, it is sufficient to create the metafile on the local disk. 43

52 Chapter 4 Functions of ReplicationControl SQL Option Restore the database file on the replication (RV or SV) created with snapshot backup by using the metafile created at the time. Attempting to restore the database file by using a metafile created at a different timing may cause an error. When performing a snapshot backup, note the following concerning the creation of meta: (1) Metafile (-metafile) specification Overwritten when the specified file already exists. If there is no folder for (or path to) the specified file name, the command fails to create meta, and the snapshot backup command results in an abnormal end. If there is a folder with the same name as the specified file name, the command fails to create meta, and the snapshot backup command results in an abnormal end. (2) Meta-folder (-metafolder) specification When the specified folder already exists, all in the folder are deleted and meta are created. If there is no folder for (or path to) the specified folder name, the command fails to create meta, and the snapshot backup command results in an abnormal end. If there is a file with the same name as the specified file name, the command fails to create meta, and the snapshot backup command results in an abnormal end. 4.4 Access Restriction It is essential that login users who connect to Microsoft SQL Server must have the right to access all databases. ReplicationControl or SnapControl commands must be executed by a user belonging to the Administrators group must. Therefore, the ReplicationControl SQL Option must also be executed by a user belonging to the Administrators group. Since User Account Control (UAC) is enabled by default in Windows Server 2008, to log on as a user other than the user belonging to the OS local system administrators group and execute these commands, the logon user must be elevated to the system administrator in advance. Therefore, to execute a command by using a command prompt, select [Run as administrator] and start the command prompt. 44

53 Chapter 5 Installation and Setting of ReplicationControl SQL Option Chapter 5 Installation and Setting of ReplicationControl SQL Option This chapter explains the work necessary to install the ReplicationControl SQL Option. (1) Operating Environment Operating system For details on the OSs supporting this software, refer to the installation guide attached to this software. Required free disk space For details on the free disk space required for installing this software, refer to the installation guide attached to this software. Required memory For details on the memory capacity required for using this software, refer to the installation guide attached to this software. (2) Installation For details on the installation procedure, refer to the installation guide attached to this software. (3) Setting Operating Environment When you use ReplicationControl SQL Option, you can use the replication operation option setting file or snapshot function option setting file to set various operations at command execution. For details on the replication operation option setting file and the snapshot function option setting file, refer to the ControlCommand Command Reference. (4) Uninstallation For details on the uninstallation procedure, refer to the installation guide attached to this software. (5) Update To update the ReplicationControl SQL Option software, uninstall the existing software, and then install the new software. For details on the installation and uninstallation procedure, refer to the installation guide attached to this software. 45

54 Chapter 5 Installation and Setting of ReplicationControl SQL Option 5.1 Setting of ReplicationControl SQL Option When constructing a system to apply the Microsoft SQL Server database and the ReplicationControl SQL Option, be sure to refer to Chapter 3 and the Microsoft SQL Server Books Online. Then design the database configuration before starting construction. If the configuration is not appropriate for the ReplicationControl SQL Option, executing this function will end in error Logical Disk Preparation Logical disks are designed and constructed to allocate them to the database to which ReplicationControl SQL Option is applied. For details, refer to the Data Replication User s Manual (Installation and Operation Guide for Windows) or Snapshot User s Manual (Installation and Operation Guide for Windows). Note that the types and configurations of available logical disks and volumes are defined Database Environment Creation The environment is constructed in the same way as the database for an ordinary Microsoft SQL Server. For details, refer to the Microsoft SQL Server Books Online. For details on database environment construction, refer to Appendix A Notes Disk Assignment for Meta The ReplicationControl SQL Option uses meta to control the snapshot. It is necessary to provide a metafile for each backup database. Meta are allocated to a different disk from the disk to which the database is allocated. In accordance with the description in 4.3, Metafile Handling, select and prepare a disk to store. When allocating to a shared disk Uses a shared disk on the network computer. Multiple meta can be allocated to a shared disk. When allocating to MV or BV Creates a metafile on the MV or BV to back it up using the data replication function or snapshot function. You cannot create a metafile on the MV or BV to which the database targeted for operation is allocated. Prepare an MV or BV other than the MV or BV to which the database is allocated. Multiple meta can be allocated to one MV or BV. When allocating to a local disk Uses a local disk on the application server. It is all right to allocate multiple meta to a local disk. 46

55 Chapter 5 Installation and Setting of ReplicationControl SQL Option 5.2 Setting of Microsoft SQL Server To use many databases in parallel, it is necessary to change the Microsoft SQL Server settings in advance. If the settings are not appropriate, when the ReplicationControl SQL Option function is executed, an error occurs resulting in an abnormal end Change of Maximum Number of Worker Threads When the snapshot backup and snapshot restoration have been executed, worker threads are consumed in the Microsoft SQL Server. Especially when snapshot backups are simultaneously performed on multiple databases, the number of worker threads consumed increases in proportion to the number of databases. When all available worker threads have been used up and new worker threads cannot be created, the ReplicationControl SQL Option results in an abnormal end, displaying the following message. In the message, the error message number (in decimal), notified by the Microsoft SQL Server, is output. A number such as 3627 and 11 is shown when all available worker threads have been used up. ism13689: Failed to snapshot execution. code=error message number Therefore, when snapshot backups are simultaneously performed on many databases, in advance, it is necessary to change the maximum number of worker threads as necessary. Indication of the maximum number of the worker threads = Default + (The number of databases on which snapshot backups are simultaneously performed 2) The default number of worker threads differs depending on the environment of the Microsoft SQL Server in use. So, refer to the Microsoft SQL Server Books Online for details. Also, those worker threads are also consumed by other processes that use the Microsoft SQL Server. Therefore, when processes other than the ReplicationControl SQL Option also use the Microsoft SQL Server, it is necessary to take into consideration the number of worker threads consumed by those processes when setting the maximum number of worker threads. Change the maximum number of worker threads according to the following procedure: (1) Start Enterprise Manager or SQL Server Management Studio from the Microsoft SQL Server s menu. (2) Select the target instance and display the properties. (3) For Enterprise Manager, select the [Processor] tab and change the maximum number of worker threads controlled by the processor. For SQL Server Management Studio, select the [Processors] node and change the maximum number of worker threads for the thread. (4) Resume the instance service. 47

56 Chapter 6 Operations of ReplicationControl SQL Option Chapter 6 Operations of ReplicationControl SQL Option This chapter describes the basic operational procedure based on using data replications and snapshot. When designing an actual operational procedure, adjust the design to the system configuration and the recovery requirements. 6.1 Data Replication Operations Tape Backup and Recovery This section explains operations of saving data from snapshot backups to tape media using the backup server without suspending operations by using ReplicationControl SQL Option. In this procedure, the MV and RV are operated in the separated state. Immediately before backup operation, the data on the MV is replicated to the RV to synchronize them. This procedure has the advantage of being able to use the RV as disk backups. Backup Master Volume (MV) Replication Volume (RV) snap1 D: Log Log D: Application Server Master Operations E: MV_Log1 Data RV_Log1 Data E: Backup Server Backup Tape Drive snap2 F: Data MV_Data Log MV_Log2 Data RV_Data Log RV_Log2 F: G: Meta Shared Disk Figure 6-1 Tape Backup 48

57 Chapter 6 Operations of ReplicationControl SQL Option The following explains the procedure for executing snapshot backup for multiple databases at a time. Logical disks to which the backup target databases are allocated have been paired and operations are running. Meta are allocated to a shared disk and can be accessed from both the application and backup servers. In this example, databases are named snap1 and snap2, the data are allocated to the same logical disk, and the log are allocated to different logical disks respectively. When the database is encrypted with transparent data encryption, back up the database master key and certificate, and store and manage them at another safe place. Workflow of the tape backup example <<Application server>> Step 1. Prepare a database operation file and replication operation file <<Backup server>> Step 2. Flush the file systems of the RV Step 3. Unmount the RV Step 4. Execute replicate for synchronization Step 5. Execute snapshot backup Step 6. Mount the RV Step 7. Back up data to a tape Step 8. Create transaction log backups 49

58 Chapter 6 Operations of ReplicationControl SQL Option Step 1. Prepare a database operation file and replication operation file (application server) Prepare a database operation file and replication operation file. For details on the database operation file and replication operation file, refer to the ControlCommand Command Reference. Sample database operation file snap_dbfile.txt #db:db_name remod:restore_mode db:snap1 remod:norecovery db:snap2 remod:norecovery * This is a sample file to be used for both snapshot backup and snapshot recovery. The above remod option specification is ignored when snapshot backup is executed and is effective when snapshot restore is executed. * When the database is encrypted with transparent data encryption, and the database master key is also encrypted with a password, the password must be specified with the dmkpass option. Sample replication operation control file snap_pairfile.txt #MV_type:MV_name RV_type:RV_name ld:mv_data ld:rv_data ld:mv_log1 ld:rv_log1 ld:mv_log2 ld:rv_log2 Step 2. Flush the file systems of the backup target disks (RV) (backup server) Flush the file systems of the backup target disks (RV) on the backup server and write the data in the file system buffer that is not yet written to the disks. ismrc_flush -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ ismrc_flush -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ ismrc_flush -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ Step 3. Unmount the backup target disks (RV) (backup server) Unmount the backup target disks (RV) on the backup server and cancel the associations between the volume and file systems. When the backup target disks (RV) are unmounted, the drive letter given to each disk is automatically removed. * For this operation, set NOTREADY for the UMOUNT_RVACC parameter in the replication operation option setting file to set RV automount disabling. ismrc_umount -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ ismrc_umount -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ ismrc_umount -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ 50

59 Chapter 6 Operations of ReplicationControl SQL Option Step 4. Execute replicate for synchronization (application server) Execute replicate to synchronize the MV and RV associated with the backup target disks. ismrc_replicate -file snap_pairfile.txt -wait Step 5. Execute snapshot backup (application server) Execute snapshot backup. When using a service to monitor the disk state, stop the service before executing snapshot backup. Restart the service after executing snapshot backup. ismsql_snapshotbkup -dbfile snap_dbfile.txt -metafolder g:\metafolder -file snap_pairfile.txt -wait The MV and RV associated with the backup target disks are separated and meta are created on the shared disk. Step 6. Mount the backup target disks (RV) (backup server) Mount the backup target disks (RV) on the backup server. By specifying the mount point volume name and drive letter for each target disk to be mounted, the drive letter is automatically set again after the disk is mounted. ismrc_mount -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ -drv d: ismrc_mount -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ -drv e: ismrc_mount -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ -drv f: Step 7. Back up data to a tape (backup server) Use backup software to back up the data and log of the databases from the backup target disks (RV) to a tape. Also back up the metadata created on the shared disk to the tape. After executing snapshot backup, you may delete the transaction log extracted before the execution of snapshot backup because it becomes unnecessary. Step 8. Create transaction log backups (application server) Periodically create transaction log backups until the next execution of snapshot backup. With these backups, you can restore the databases to the status when the transaction log backups were created during recovery performed if an error occurs. 51

60 Chapter 6 Operations of ReplicationControl SQL Option Recovery The following explains the procedure for performing recovery using database backups obtained by executing snapshot backup. This example assumes that an error occurs in a data file on the MV. The tape contains backups and meta obtained by executing snapshot backup. In this example, databases are named snap1 and snap2, the data are allocated to the same logical disk, and the log are allocated to different logical disks respectively. Master Volume (MV) Replication Volume (RV) snap1 D: Log Log D: Application Server Master Operations E: MV_Log1 Error Data RV_Log1 Data E: Backup Server Backup Tape Drive snap2 Data MV_Data Data RV_Data F: F: Log Log MV_Log2 RV_Log2 G: Meta Shared Disk Figure 6-2 Recovery Using Tape Backup If an error occurs on the logical disk to which the data of database snap1 and snap2 are allocated, but not on the logical disk to which the log of each database are allocated, you can use Microsoft SQL Server to create transaction log backups. In this case, you can restore the databases to the status immediately before the error occurred by applying the created transaction log backups after executing snapshot restore. When you periodically create transaction log backups, if an error occurs on the logical disk to which log are allocated, you can restore the databases to the status to a point until the latest transaction log backups were created. If transaction log backups are lost, you cannot restore a database to the status when the transaction log backups were created. For this reason, save the transaction log backups on a highly reliable disk. 52

61 Chapter 6 Operations of ReplicationControl SQL Option Workflow of the example of recovery using tape backup <<Application server>> Step 1. Prepare a database operation file and replication operation file <<Backup server>> Step 2. Stop the operations Step 3. Create transaction log backups Step 4. Detach the databases Step 5. Restore data from the tape Step 6. Flush the file systems of the RV Step 7. Unmount the RV Step 8. Execute snapshot restore Step 9. Apply the transaction log Step 10. Restart the operations Step 11. Confirm the completion of restoration Step 12. Mount the RV Step 1. Prepare a database operation file and replication operation file (application server) Use the database operation file and replication operation file prepared in step 1 of the backup procedure. Step 2. Stop the operations (application server) Stop the operations. 53

62 Chapter 6 Operations of ReplicationControl SQL Option Step 3. Create transaction log backups (application server) If an error occurs in a data file, create transaction log backups. When the database is encrypted with transparent data encryption, and the database master key is also encrypted with a password, decrypt them before backing up the transaction log. T-SQL : USE MASTER;OPEN MASTER KEY DECRYPTION BY PASSWORD = 'sfj5300osdvdgwdfkli7' In this example, transaction log backups are created on the local disk (H drive) on the application server. T-SQL : BACKUP LOG snap1 TO DISK = h:\sql\backups_snap1\trbl.dat WITH NO_TRUNCATE T-SQL : BACKUP LOG snap2 TO DISK = h:\sql\backups_snap2\trbl.dat WITH NO_TRUNCATE In Microsoft SQL Server 2008 or later, you can compress and back up the transaction log file by specifying the COMPRESSION option. T-SQL : BACKUP LOG snap1 TO DISK = 'H: sql backups_snap1 trbl.dat' WITH NO_TRUNCATE,COMPRESSION T-SQL : BACKUP LOG snap2 TO DISK = 'H: sql backups_snap2 trbl.dat' WITH NO_TRUNCATE,COMPRESSION Step 4. Detach the databases (application server) Detach and disconnect the databases in which the error occurred. T-SQL : ALTER DATABASE snap1 SET SINGLE_USER WITH ROLLBACK IMMEDIATE T-SQL : EXEC sp_detach_db snap1 T-SQL : ALTER DATABASE snap2 SET SINGLE_USER WITH ROLLBACK IMMEDIATE T-SQL : EXEC sp_detach_db snap2 Step 5. Restore data from the tape (backup server) Use backup software to restore data from the tape to the RV. Also restore the meta on the shared disk. Step 6. Flush the file systems of the recovery target disks (RV) (backup server) Flush the file systems of the backup target disks (RV) on the backup server and write the data in the file system buffer that is not yet written to the disks. ismrc_flush -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ ismrc_flush -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ ismrc_flush -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ 54

63 Chapter 6 Operations of ReplicationControl SQL Option Step 7. Unmount the recovery target disks (RV) (backup server) Unmount the backup target disks (RV) on the backup server and cancel the associations between the volume and file systems. When the recovery target disks (RV) are unmounted, the drive letter given to each disk is automatically removed. * For this operation, set NOTREADY for the UMOUNT_RVACC parameter in the replication operation option setting file to set RV automount disabling. ismrc_umount -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ ismrc_umount -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ ismrc_umount -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ Step 8. Execute snapshot restore (application server) Execute snapshot restore with specifying RV protection (-mode protect). Restore the databases in the state in which the transaction log can be applied (norecovery). When using a service to monitor the disk state, stop the service before executing snapshot restore. Restart the service after executing snapshot restore. ismsql_snapshotrst -dbfile snap_dbfile.txt -metafolder g:\metafolder -mode protect The databases are restored in the state in which the transaction log can be applied. When snapshot restore is executed with RV protection specified, the MV and RV are automatically separated at the completion of the restoration. For this reason, it is not necessary to execute separate. Step 9. Apply the transaction log (application server) Apply the transaction log to the databases. Apply the backups of the transaction log extracted after the execution of snapshot backup in the order in which they were created. In Microsoft SQL Server 2008 or later, you can directly apply the transaction log compressed at backup, without decompressing, to the database. When the database is encrypted with transparent data encryption, and the database master key is also encrypted with a password, decrypt them before applying the transaction log to the database. T-SQL : USE MASTER;OPEN MASTER KEY DECRYPTION BY PASSWORD = 'sfj5300osdvdgwdfkli7' In this example, the backups of the transaction log are stored on the local disk (H drive) on the application server. T-SQL : RESTORE LOG snap1 FROM DISK = h:\sql\backups_snap1\backup01.dat WITH NORECOVERY T-SQL : RESTORE LOG snap2 FROM DISK = h:\sql\backups_snap2\backup01.dat WITH NORECOVERY : T-SQL : RESTORE LOG snap1 FROM DISK = h:\sql\backups_snap1\trbl.dat WITH NORECOVERY 55

64 Chapter 6 Operations of ReplicationControl SQL Option T-SQL : RESTORE LOG snap2 FROM DISK = h:\sql\backups_snap2\trbl.dat WITH NORECOVERY After applying all transaction log backups, restore the databases. T-SQL : RESTORE DATABASE snap1 WITH RECOVERY T-SQL : RESTORE DATABASE snap2 WITH RECOVERY Step 10. Restart the operations (application server) Make sure that Microsoft SQL Server and the databases are in the normal state, then restart the operations. Step 11. Confirm the completion of restoration (backup server) Wait for the completion of separation performed at the completion of the restoration. ismrc_wait -cond sep -file snap_pairfile.txt Step 12. Mount the backup target disks (RV) (backup server) Mount the backup target disks (RV) on the backup server. By specifying the mount point volume name and drive letter for each target disk to be mounted, the drive letter is automatically set again after the disk is mounted. ismrc_mount -mvol \\?\Volume{d9ea193d-9d59-11d6-ab08-806d f}\ -drv d: ismrc_mount -mvol \\?\Volume{d9ea193e-9d59-11d6-ab08-806d f}\ -drv e: ismrc_mount -mvol \\?\Volume{d9ea193c-9d59-11d6-ab08-806d f}\ -drv f: After recovery, back up data in the database environment as required. 56

65 Chapter 6 Operations of ReplicationControl SQL Option Disk Backup and Recovery The use of the logical disks themselves of RVs as a backup device makes the backup server unnecessary and enables a series of operations only with the application server. This mode requires the switching between multiple RVs to use them, however. Securing data consistency on the application server is the same as when tape backup is performed. Backup Master Volume (MV) Replication Volumes (RVs) D: snap1 Log Log MV_Log1 RV1_Log1 Log RV2_Log1 Application Server E: Master Operations Data Data Data snap2 Data MV_Data Data RV1_Data Data RV2_Data F: Log Log MV_Log2 G: Meta MV_Meta RV1_Log2 Meta RV1_Meta Log RV2_Log2 Meta RV2_Meta Figure 6-3 Disk Backup The following explains the procedure for switching between multiple RVs and executing snapshot backup. Logical disks to which the backup target databases are allocated have been paired and operations are running. Meta are created on the logical disk (MV) in the pair setting state, and after the execution of snapshot backup, they are backed up on the RVs. In this example, databases are named snap1 and snap2, the data are allocated to the same logical disk, and the log are allocated to different logical disks respectively. When the database is encrypted with transparent data encryption, back up the database master key and certificate, and store and manage them at another safe place. 57

66 Chapter 6 Operations of ReplicationControl SQL Option Workflow of the disk backup example <<Application server>> Step 1. Prepare a database operation file and replication operation <<Backup server>> Step 2. Execute replicate for synchronization Step 3. Execute snapshot backup Step 4. Flush the file system of the MV Step 5. Unmount the MV Step 6. Backup the meta Step 7. Mount the MV Step 8. Switch between operation target pairs Step 9. Create transaction log backups Step 1. Prepare a database operation file and replication operation (application server) Prepare a database operation file and replication operation. For details on the database operation file and replication operation file, refer to the ControlCommand Command Reference. Sample database operation file snap_dbfile.txt #db:db_name remod:restore_mode db:snap1 remod:norecovery db:snap2 remod:norecovery * This is a sample file to be used for both snapshot backup and snapshot recovery. The above remod option specification is ignored when snapshot backup is executed and is effective when snapshot restore is executed. * When the database is encrypted with transparent data encryption, and the database master key is also encrypted with a password, the password must be specified with the dmkpass option. 58

67 Chapter 6 Operations of ReplicationControl SQL Option Sample replication operation snap_pairfile1.txt #MV_type:MV_name RV_type:RV_name ld:mv_data ld:rv1_data ld:mv_log1 ld:rv1_log1 ld:mv_log2 ld:rv1_log2 meta_pairfile1.txt #MV_type:MV_name RV_type:RV_name ld:mv_meta ld:rv1_meta snap_pairfile2.txt #MV_type:MV_name RV_type:RV_name ld:mv_data ld:rv2_data ld:mv_log1 ld:rv2_log1 ld:mv_log2 ld:rv2_log2 meta_pairfile2.txt #MV_type:MV_name RV_type:RV_name ld:mv_meta ld:rv2_meta Step 2. Execute replicate for synchronization (application server) Execute replicate for the current operation target pairs to synchronize the MV and RV associated with the backup target disks. ismrc_replicate -file snap_pairfile1.txt -wait ismrc_replicate -file meta_pairfile1.txt -wait Step 3. Execute snapshot backup (application server) Execute snapshot backup for the operation target pairs associated with the backup target disks to which the databases are allocated. When using a service to monitor the disk state, stop the service before executing snapshot backup. ismsql_snapshotbkup -dbfile snap_dbfile.txt -metafolder g:\metafolder -file snap_pairfile1.txt -wait The MV and RV associated with the backup target disks to which the databases are allocated are separated and meta are created on the metafile storage disk. 59

68 Chapter 6 Operations of ReplicationControl SQL Option Step 4. Flush the file system of the metafile storage disk (MV) (application server) Flush the file system of the metafile storage disk (MV) and write the data in the file system buffer that is not yet written to the disk. ismrc_flush -drv g: Step 5. Unmount the metafile storage disk (MV) (application server) Unmount the metafile storage disk (MV) and cancel the associations between the volume and file system. ismrc_umount -drv g: Step 6. Back up the meta (application server) Execute separate for the operation target pair associated with the metafile storage disk. ismrc_separate -file meta_pairfile1.txt -wait The MV and RV associated with the metafile storage disk are separated. Step 7. Mount the metafile storage disk (MV) (application server) Mount the metafile storage disk (MV). ismrc_mount -drv g: Restart any service if it has been stopped in step 3. Step 8. Switch between operation target pairs (application server) To switch between operation target pairs, rename the replication operation to cycle pairs. The file names of the replication operation containing the operation target pair are always snap_pairfile.txt and meta_pairfile1.txt. move snap_pairfile1.txt snap_temp.txt move meta_pairfile1.txt meta_temp.txt move snap_pairfile2.txt snap_pairfile1.txt move meta_pairfile2.txt meta_pairfile1.txt move snap_temp.txt snap_pairfile2.txt move meta_temp.txt meta_pairfile2.txt 60

69 Chapter 6 Operations of ReplicationControl SQL Option Figure 6-4 Switching between Disk Backup Target Pairs Step 9. Create transaction log backups (application server) Periodically create transaction log backups until the next execution of snapshot backup. With these backups, you can restore the databases to the status when the transaction log backups were created during recovery performed if an error occurs. To execute backup after that, perform operations from step 1. 61