DB2 10 LUW: Advanced Features

Transcription

1 DB2 10 LUW: Advanced Features Information Management 2012 IBM Corporation

2 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security - LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

3 High Availability and Disaster Recovery - HADR High Availability: keeps the data available whenever required by eliminating or minimizing downtime. Commons solutions for HA: Advanced storage scenario (like split mirroring) Database partitioning with clustering software Data replication (SQL or Q replication) Disaster Recovery: protect against data loss caused by unrecoverable failure. Common solutions for DR: Backup and Restore Log shipping IBM Corporation

4 HADR Concept HADR is a database replication feature that provides a high availability solution for both partial and complete site failures. HADR is based on a pair (primary and standby) of databases. In case of failure, the standby database can take over the workload Primary Handles all client connections and processes transactions Continuously ships transaction logs to the standby over the network Standby Kept in sync with primary by applying received transaction logs tx tx Network Connection HADR Keeps the two servers in sync Primary Server Standby Server IBM Corporation

5 HADR Capabilities High Availability Automatic Failover with Tivoli System Automation Automatic Client Reroute Primary and/or Standby Failure (planned or unplanned) Reads-on Standby Offloading read transactions (and log replication) from the primary server to standby servers Multiple Standby NEW IN DB2 10 Local standby for high availability Remote standby for disaster recovery City B Read-only session 5 Transactions City A Server 1 Database A Primary HADR Read / Write session 1 Read / Write session 2 Read-only session 4 Transactions Server 3 Database A Standby HADR Server 2 Database A Standby HADR Read--only session IBM Corporation

6 HADR Scope of Action HADR replication takes place at the database level Server 1 Server 2 Database A Primary HADR TCP/IP Standby HADR Database A Database C Database B Database D Database E Standby HADR TCP/IP Primary HADR Database E IBM Corporation

7 HADR Synchronization Modes Synchronous Greatest protection against transaction loss A higher cost of transaction response time Near Synchronous (default) Better performance than Synchronous mode no waiting for the log buffer to be written to disk on standby Asynchronous If latency of the network between the two servers is too great Super Asynchronous Better response time during network interruptions or congestion Primary Log shipping Standby HADR send buffer Log Writer Asynchronous HADR receive buffer Log Writer Near-synchronous Log Super Asynchronous Synchronous Log IBM Corporation

8 Configuring Reads on Standby (RoS) Perform read-only operations on HADR standby database Run concurrent read-only workloads offloading reporting, DSS/BI workloads to standby with minimal impact Enabled by DB2_HADR_ROS registry variable During failover, DB2 seamlessly turns the read-on-standby to a primary read/write server Restrictions Only Uncommitted Read (UR) isolation level is supported on the standby Workload manager DDL statements on the primary are replayed on the standby, but they will not be effective on the standby Some DB2 features and objects are not supported CGTTs, DGTTs, not logged initially tables, not-inlined LOBs and XML, LONG VARCHAR, LONG GRAPHIC, STMM Server 1 Read/Write session 1 Read/Write session 2 Server 2 Read session 3 Database A Primary HADR Writes for session 1 Writes for session 2 Standby Database A (with RoS) IBM Corporation HADR

9 HADR Configuration HADR Delayed Replay HADR_REPLAY_DELAY - New configuration parameter which will control how far behind (in seconds) the standby will intentionally remain at all times Very useful to prevent data loss due to errant transactions. E.g.: You can delay the HADR replication by 24h, creating a safe point-of-restore for unwanted transactions. Max value is seconds Can only be set on a standby database A TAKEOVER command on a standby with replay delay enabled will fail A standby database with replay delay enabled cannot take over as the new primary until you have disabled the delayed replay The delayed replay feature is supported only in SUPERASYNC mode New in DB2 10 Database A Primary HADR Transaction Logs Standby HADR Database A Delay Logs IBM Corporation

10 HADR Fine Tuning a Standby Log Spooling The log spooling feature allows the standby to spool log records arriving from the primary Decouple log replay on the standby from the receiving of log data from the primary Logs will be spooled in the standby DB s active log path No data loss Reduces impact on primary database caused by slow log replay on standby. E.g.: When intensive operations like REORGs are replayed on the standby. Set through a new DB CFG parameter: HADR_SPOOL_LIMIT Specify maximum amount of disk space to use for log spooling Value of 0 disables spooling (default) Value of -1 defines the spool to be unlimited (limited by file system free space) Maintenance: Log files will be automatically deleted when they are no longer required Takeover may take longer as the spool must be drained before takeover completes IBM Corporation

11 Multiple Standby Overview Primary super async mode only Auxiliary Standby Support for up to two Auxiliary Standbys (AS) - AS supports super async mode only - No automated takeover supported - Always feed from the current primary - Can be added dynamically any sync mode Principal Standby Auxiliary Standby Principal Standby (PS) equivalent to standby today - PS supports any sync mode - Can automate takeover using integrated TSA IBM Corporation

12 Oracle Data Guard vs. DB2 HADR Oracle has the same modes as DB2's HADR. Oracle Mode Maximum Performance Maximum Availability Maximum Protection Not available Feature Availability Comparison DB2 Mode DB2 asynchronous DB2 near synchronous DB2 synchronous DB2 super asynchronous Oracle Express Not available DB2 Express-C Not available Oracle Standard One Not available DB2 Express FTL DB2 Express Included Purchase Oracle Standard Not available DB2 Workgroup Included Oracle Enterprise Purchase DB2 Enterprise / Advanced Enterprise Included IBM Corporation

13 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

14 DB2 purescale Designed for OLTP Extreme Capacity Buy only what you need, add capacity as your needs grow Handle key capacity spikes with pay by the day pricing Application Transparency Easy to implement, easy to grow Continuous Availability 24x7 availability so key database systems never go down, even if multiple servers fail IBM Corporation

15 Supported Platforms OR POWER6 POWER7 SLES 10 SP3 SLES 11 SP1 x3650 M3 RHEL 5.6 RHEL 6.1 x3850 X5 x3690 X5 BladeCenter H22 GPFS compatible storage IBM Corporation

16 Oracle Real Application Clusters (RAC) Instance 1 fails Instance 1 Instance 2 Instance 3 GRD GRD GRD Must read log and lock pages before freeze is lifted. I/O Requests are Frozen x x x x x x Recovery instance locks pages that need recovery Distributed locking in RAC results in higher overhead and lower scalability Oracle RAC best practices recommends Fewer rows per page (to avoid hot pages) Partition database to avoid hot pages Partition application to get some level of scalability All of these result in higher management and development costs IBM Corporation

17 DB2 purescale vs. Oracle RAC DB2 Provides central locking and buffer pool (cache) management Utilize RDMA to eliminate interrupts for messaging and allow direct memory access Duplexed Cluster Facility (CF) to ensure scalability and availability CF software bundled with purescale, no separate license required ORACLE No centralized locking or buffer cache management Requires messaging between Oracle RAC nodes, can lead to scalability issues as cluster grows in size In RAC, messaging between members uses standard IP (does not take advantage of RDMA) IBM Corporation

18 The Result 128 Members 84% Scalability 112 Members 89% Scalability 2, 4 and 8 Members Over 95% Scalability 64 Members 95% Scalability 88 Members 90% Scalability 16 Members Over 95% Scalability 32 Members Over 95% Scalability IBM Corporation

19 Enhancements in DB2 10 Added support to Range-partitioned tables All roll-in and roll-out operations ADD/ATTACH/DETACH PARTITION ALTER TABLE SALES ATTACH PARTITION pt1 on SALES Asynchronous partition attach will start This is only run on a single member It may be different from the member that issued the attach Leverage Partitioned indexes and Partition REORGs DB2 Workload Manager now available with DB2 purescale Using a split mirror as a backup image Added support to SET WRITE operations New CURRENT MEMBER default value improves DB2 purescale performance This member information can then be used to range partition a table or an index, and therefore reduce database contention IBM Corporation

20 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

21 DB2 Database Partition Feature (DPF) A partitioned database environment is a database installation that supports the distribution of data across database partitions. A database partition is a part of a database that consists of its own data, indexes, configuration files, and transaction logs Data access across partitions is transparent to applications Leverages the power of multiple CPUs on multiple physical machines to efficiently process complex queries like those find in OLAP or data mining applications Improved Performance Query Optimization Split Workload Parallelism Benefits Shared Nothing Architecture DPF feature is available as an add-on feature for InfoSphere Warehouse Scalability IBM Corporation

22 DPF Configurations Single Logical Partition on Single Physical Partition Multiple Logical Partitions on Single Physical Partition Single Logical Partition on Multiple Physical Partitions Multiple Logical Partitions on Multiple Physical Partitions IBM Corporation 22

23 DPF Node Configuration (db2nodes.cfg) All database partitions that are specified in the database partition configuration file (db2nodes.cfg) are created with the database DB partition number Unique database partition ID Server hostname Machine s name or IP address Server logical-port Logical partition ID within a machine Partition Server Name Logical Port 0 Server A 0 1 Server A 1 2 Server B 0 3 Server B 1 4 Server C 0 5 Server C IBM Corporation

24 Table Partitioning Allows a single logical table to be broken up into multiple separate physical storage objects (a.k.a. data partitions) Up to 32K data partitions Each partition defines a range of values A partition will only contain rows that match its range of values Parallel table scans and index scans IBM Corporation

25 Table Partitioning - Features Roll-in / Roll-out: ALTER TABLE ATTACH / DETACH Incorporates / exclude an existing table as a new range Use SET INTEGRITY to validate data and maintain NEW IN DB2 10 indexes after ATTACH operation Use SET INTEGRITY IMMEDIATE UNCHECKED for quicker data availability Minimal interruption to other queries accessing table No data movement Fast Operation ATTACH or ADD completes immediately, allowing new partition to be accessible sooner for non-rr queries Table Partition Elimination Ability to determine that only a subset of the data partitions in a table are necessary to answer a query SELECT * FROM PAYMENTS WHERE paydate BETWEEN '02/03/2009' AND '30/05/2009' Jul Aug Sep 2009 Oct Nov Dec IBM Corporation

26 Indexes on Partitioned Tables Without table partitioning, all indexes for a particular table are stored in the same storage object by default With table partitioning, each index can be placed on its own table space Including MDC (aka block) indexes Benefits Improved data roll-in and roll-out performance More granular control of index placement Improved performance on dropping and creating online indexes Space is freed the moment an index is dropped More efficient concurrent access to index data for the table due to reduced I/O contention i1 Global Index Payments Partition 1 Partition 2 Partition 3 Partition 4 Jan Feb Mar Apr May Jun Jul Ago Sep Oct Nov Dec i2 i2 i2 i2 Partitioned Index IBM Corporation

27 Multi-Dimensional Clustering Blocks: consecutive set of pages on the disk. Block Indexes: indexes that point to an entire block of pages Cells: portion of the table that contains rows having the same unique set of dimension values Dimension: axis along which data is organized in an MDC table Good candidates: Columns referenced by ORDER BY and GROUP BY clauses Columns used for range, equality, and IN predicates Roll-in or roll-out of data Slice: portion of the table that contains all the rows that have a specific value for one of the dimensions IBM Corporation

28 Combining All Features Single large table without partitioning SELECT * FROM GRAPHICS where Each I/O can pick up many unrelated records that happen to reside on the same page IBM Corporation

29 Combining All Features SELECT * FROM GRAPHICS where Using Database Partitioning Take the advantage of the query parallelism provided by DB2 DPF attacking each partition in parallel IBM Corporation

30 Combining All Features Using Database Partitioning and Table Partitioning SELECT * FROM GRAPHICS where All data in the same user-defined range is consolidated in the same data partition. DB2 can limit query search to only the suitable partitions. Savings in I/O operations IBM Corporation

31 Combining All Features Using Database Partitioning, Table Partitioning, and MDC SELECT * FROM GRAPHICS where Even less I/O is performed to retrieve the records of interest. DB2 can retrieve more matching pages as data is clustered according to the dimensions IBM Corporation

32 Insert Time Clustering (ITC) Tables New in DB2 10 Clusters data inserted at a similar time using an implicitly created virtual dimension Virtual dimension cannot be manipulated Cannot be used if table is a typed table Manage database size more effectively without manual intervention Very similar to MDC tables. It uses block based allocation and block indexes. DB2 CREATE TABLE T1(c1 int, c2 char(100), ) IN TABLESPACE1 ORGANIZE BY INSERT TIME; DB2 CREATE INDEX INX1 ON T1(C1); INSERT TIME IBM Corporation

33 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

34 Temporal Tables What are They? Temporal tables allow you to associate time-based state information to your data Managed by DB2, not application logic DB2 s temporal table allow Tracking of data changes (versioning) Support for custom business data storage and manipulation A combination of the two Benefits Increase business insight Easily report on data at a given point in time in the past, present, or future Lower costs Increase developer productivity less coding by offloading time management to DB2 Simpler code means reduced maintenance costs Allow you to reduce compliance efforts and better track data changes Better manage data Standards-based technology provides consistency and data quality across the enterprise Seamless and deep integration with all database features such as partitioning and compression IBM Corporation

35 Temporal Tables Types Built into DB2 automatic and transparent Three types of temporal tables System-period temporal tables (STTs) DB2 automatically maintains historical versions of the rows in the history table You can query the past state of your data Example Employees who have left the company Application-period temporal tables (ATTs) You assign a date range to each row, indicating the period when the data is valid in the real world Valid periods can be in the past, present, or future Example Insurance policy valid from Jan 1 to June 30 4% interest rate is effective from Nov 1 to 20, and on Nov 21 the effective rate will increase to 5% Bitemporal Tables Combination of STT and ATT Keep application-based period information as well as system-based historical information IBM Corporation

36 System-Period Temporal Tables (STTs) Behavior DB2 automatically migrates rows from base table to history table as changes occur, updating the three columns as required All management of base and history tables is automatic and transparent. Queries only reference the base table, and DB2 will transparently access the history table as needed TRAVEL System validity period (inclusive, exclusive) trip_name destination departure_date price sys_start sys_end tx_start_id Amazonia Brazil 10/28/ /01/ /30/ /01/2011 Lake Tahoe, CA 01/01/ /15/2011 Ski Heavenly Valley California 03/01/ /15/ /01/ /30/9999 SYS_START, SYS_END and tx_start columns are inserted by DB2, not the application. For simplicity, they are represented here as DATEs, rather than TIMESTAMPs UPDATE travel SET destination = 'Lake Tahoe, CA' WHERE trip_name = 'Ski Heavenly Valley ; TRAVEL_HISTORY trip_name destination departure_date price sys_start sys_end Ski Heavenly Valley California 03/01/ /01/ /15/2011 Current Date = February 15, 2011 tx_start_id 01/01/ IBM Corporation 36

37 Application-Period Temporal Tables (ATTs) Behavior: Each row has a pair of timestamp or date columns, stored by the application Unlike System-period temporal tables (STTs), no separate history table is required Benefits DB2 adds, splits, or deletes rows transparently as needed Can be used to model data in the past, present and future Constraints can be automatically enforced to disallow overlapping validity periods UPDATE travel FOR PORTION OF BUSINESS_TIME FROM '06/01/2011' TO '07/01/2011' SET price = WHERE trip_name = 'Manu Wilderness'; **Application-period time entries are independent of the current date Current Date = May 15, 2011 BUSINESS_TIME period (inclusive, exclusive) trip_name destination departure_date price bus_start bus_end Manu Wilderness Peru 08/02/ /01/ /01/ /01/2011 Manu Wilderness Peru 08/02/ /01/ /01/2011 Manu Wilderness Peru 08/02/ /01/ /01/2012 DB2 inserted 2 rows and updated 1 row IBM Corporation

38 Bitemporal Tables Combine application-period (ATT) and system-period (STT) capabilities Every row has a pair of: timestamps set by DB2 (SYSTEM_TIME period) business timestamps or date columns (BUSINESS_TIME period) set by the application trip_name destination departure_date price bus_start bus_end sys_start sys_end Alligator Swamp Louisiana 02/15/ /01/ /16/ /01/ /30/9999 Query using both application-period time and system-period time Example: what trips were offered on June 20, 2011, as recorded in the database on May 10, 2011? SELECT trip_name, destination FROM TRAVEL FOR BUSINESS_TIME AS OF '06/20/2011' FOR SYSTEM_TIME AS OF ' '; INSERT/UPDATE/DELETE behavior is similar to ATTs (rows inserted/split/deleted as required) UPDATE and DELETE cause automatic insertion into the corresponding history table SELECT will go to the history table as needed to get rows IBM Corporation

39 More on Temporal Tables Views may be defined on system-period temporal tables (base and history), application-period temporal tables, or bitemporal tables for select or update Special registers allow querying past or future points in time You can set the clock back or forward to a specific time for a given session No changes required on applications! Setting these registers allows you to query: Past point in SYSTEM_TIME Past or future point in BUSINESS_TIME DB2SET CURRENT TEMPORAL SYSTEM_TIME = CURRENT TIMESTAMP 1 YEAR DB2SET CURRENT TEMPORAL BUSINESS_TIME = ' ' Current tables can be converted to STTs or ATTs IBM Corporation

40 Oracle Time Travel Query Comparison DB2 Can limit history capture to critical tables Captures table changes at the row level Each table using system-period temporal tables has its OWN history table Technology is part of the core database engine Available with ALL Editions ORACLE Cannot limit history capture by table (only by space allocated to UNDO) Captures data changes in UNDO table space for ALL changes, Captures table changes at the data block level history maintained in a separate table space(s) Must use a different table space for each table if data separation required To guarantee a retention period requires Flashback Data Archive (aka Total Recall) option on Oracle Enterprise Edition IBM Corporation

41 Time Travel Tables summarized Temporal tables enable time travel! Temporal tables can be used as: System-period temporal tables (STTs) Managed by DB2 DB2 maintains a separate history table Application-period temporal tables (ATTs) Managed by the application Current and historical rows are all in the base table Bitemporal tables Combine System-period and Application-period temporal tables Views can be created on STTs or ATTs for select or update Special registers allow querying past or future points in time Current tables can be converted to STTs or ATTs IBM Corporation

42 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

43 Using Advanced Data Security DB2 provides ways to control data access at row / column level Set up rich security policies Prevents administrators with DATAACCESS authority from accessing all data in a database No dependency on application logic Facilitates table level multi-tenancy How can we ensure that managers only see data for their own employees? How can we ensure that a social security number column is masked out for unauthorized users? IBM Corporation

44 Two Ways to Protect your Data LBAC Label Based Access Control It is a fixed label security model designed for the US government Hierarchical access scenarios Great for large companies with well defined data and user classifications Suited for such applications as those intelligence and defense communities NEW IN DB2 10 RCAC Row and Column Access Control It is a general purpose security model No data or user classification required Best suited for commercial customers IBM Corporation

45 DB2 LBAC Label Based Access Control Access Control at the table level via traditional privileges Does the user hold the required privilege to perform the requested operation on the table? Label Based Access Control Sets security labels at the row level, column level or both How does LBAC work? Users and Objects (rows/columns) are assigned labels that are later compared to authorize access DEPT_1 DEPT_2 Employee HR Manager IBM Corporation

46 Use Case: Defining Row/ Column Labels 3 column labels: NON_SENS for non-sensitive data SENS for sensitive data CONF for confidential data 2 row labels: DEPT_1 for Department 1 DEPT_2 for Department IBM Corporation

47 What is Row and Column Access Control? RCAC is: Additional layer of data security introduced in DB2 10 Complementary to table level authorization Allows access only to subset of data useful for job task Controls access to a table at the row, column or both levels Two sets of rules, permissions for rows and masks for columns IBM Corporation

48 DB2 RCAC - Row Permission An access control rule defined by a SQL statement that describes what set of rows a user has access to. To use RCAC row permissions: Row permission name Table/view name Determines if permission will be enabled when access control is activated for table CREATE PERMISSION row_access ON patient FOR ROWS WHERE ( verify_role_for_user(session_user, PATIENT ) = 1 AND patient.userid = SESSION_USER ) OR ( verify_role_for_user(session_user, PCP ) = 1 AND patient.pcp_id = SESSION_USER ) OR ( verify_role_for_user(session_user, ACCOUNTING ) = 1 ) ENFORCED FOR ALL ACCESS ENABLE; ALTER TABLE patient ACTIVATE ROW ACCESS CONTROL; Search-condition Only users with SECADM authority can create RCAC rules Table/view name Activate row access control IBM Corporation

49 DB2 RCAC - Row Permission Example UPDATE patient SET pharmacy = codeine WHERE name = Sam Successful update statement! SIN USERID NAME ADDRESS PHARMACY ACCT_BALANCE PCP_ID MAX Max First St. hypertension LEE MIKE Mike Long St. diabetics 8.30 JAMES SAM Sam Big St. codeine LEE DOUG Doug Good St. influenza 7.68 JAMES BOB Bob 123 Some St. hypertension 9.00 LEE UPDATE patient SET pharmacy = codeine WHERE name = Doug Unsuccessful update statement! No row was found for FETCH, UPDATE or DELETE; or the result of a query is an empty table. If you can t view a row, you can t update it either IBM Corporation

50 DB2 RCAC - Column Mask A access control rule defined by a SQL CASE expression that describes what column values a user is permitted to see and under what conditions. To protect a column with an RCAC column mask Table/view name Determines if mask will be enabled when access control is activated for table 1 2 CREATE MASK acct_balance_mask ON patient FOR COLUMN acct_balance RETURN CASE WHEN verify_role_for_user(session_user, ACCOUNTING ) = 1 THEN acct_balance ELSE 0.00 END ENABLE; CREATE MASK sin_mask ON patient FOR COLUMN sin RETURN CASE WHEN verify_role_for_user(session_user, PATIENT ) = 1 THEN sin ELSE XXX XXX SUBSTR(sin,8,3) END ENABLE; ALTER TABLE patient ACTIVATE COLUMN ACCESS CONTROL; Column mask name Result of case is returned in substitute of column value Activate column access control IBM Corporation

51 DB2 RCAC - Column Mask Example 1 Accountants can see account balances. 2 Accountants cannot see SIN numbers. SELECT * FROM patient SIN USERID NAME ADDRESS PHARMACY ACCT_BALANCE PCP_ID XXX XXX 234 MAX Max First St. hypertension LEE XXX XXX 812 MIKE Mike Long St. diabetics 8.30 JAMES XXX XXX 856 SAM Sam Big St. codeine LEE XXX XXX 454 DOUG Doug Good St. influenza 7.68 JAMES XXX XXX 789 BOB Bob 123 Some St. hypertension 9.00 LEE 1 Doctors cannot see account balances. 2 Doctors cannot see SIN numbers. SELECT * FROM patient Doctors can only see the rows of their own patients (from Row Access Control) SIN USERID NAME ADDRESS PHARMACY ACCT_BALANCE PCP_ID XXX XXX 234 MAX Max First St. hypertension 0.00 LEE XXX XXX 856 SAM Sam Big St. codeine 0.00 LEE XXX XXX 789 BOB Bob 123 Some St. hypertension 0.00 LEE IBM Corporation

52 DB2 RCAC vs. Oracle DB2 Provides easy implementation via SQL statements Controls row and column access permissions via SQL Controls all DML and SELECT against a table RCAC provides flexible masking that is user configurable Full control of mask display (blanks, partial column value, etc.) ORACLE No comparative implementation via SQL Requires Virtual Private Database to provide a similar functionality More difficult to implement and manage Masking is limited to NULL display ONLY Can easily enable/disable enforcement IBM Corporation

53 Protected Tables A protected table is a table on which LBAC and RCAC is enforced Types of protected tables: Protected table with row level granularity Protected table with column level granularity Under LBAC and RCAC, DBADM and SYSADM have no inherent ability to access data within a protected table! No security label or exemption, no data access! The content of a protected table with row level granularity appears different depending on the user accessing such table You only see the rows which your LBAC credentials or your RCAC permissions allows you to see IBM Corporation

54 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

55 Storing XML: Native XML Storage Relational columns are stored in relational format (tables) Values of XML columns are stored natively (XML documents in parsed hierarchical format), encoded in UTF- 8 XML values are stored: In the purexml storage area (by default). A descriptor points from row to XML storage Or inlined in row, if enabled for XML document smaller than 32KB. No XML parsing for query evaluation! XML data can be compressed IBM Corporation

56 XML Storage: DB2 vs. Others Unstructured XML storage: XML as text Others Decomposition : XML Relational DB2 purexml XML as XML XML DOC Extract selected elements/attr. XML DOC Fixed Mapping XML DOC XML DOC XML DOC XML DOC CLOB Column Using i.e. a XML CLOB as text or VARCHAR Column XML Index Side Tables or Indexes Any sub-document level access requires XML parsing slow. (regular relational tables) Mapping prevents XML schema changes, and is often too complex. XML reconstruction is slow. Maximum flexibility and performance XML Column IBM Corporation

57 Quick Overview of purexml Capabilities Support to XQuery and SQL / XML XML schema validation Transformation with XSLT stylesheet XML indexing Compression of XML data Application development support for C, C++, COBOL, JAVA, C#, VB, PHP, PERL and SQL routines IBM Corporation

58 Improved Insert/Query Performance with Binary XML Format New in DB2 10 Improve end to end performance of client/server applications using JDBC or SQLJ interfaces Avoid conversion between XML as text and internal format (client / server) by introducing XDBX Already used in DB2 for z/os Simple to generate, simple to parse Provides improved end to end performance Transparent to applications Storage and retrieval of binary XML data requires version 4.9 or later of the IBM Data Server Driver for JDBC and SQLJ IBM Corporation

59 Agenda High Availability and Disaster Recovery HADR DB2 purescale DB2 Partitioning DB2 Temporal Tables DB2 Advanced Security LBAC & RCAC DB2 purexml DB2 Workload Management IBM Corporation

60 Workload Management in DB2 Stable and predictable execution environment Light-weight and granular Better resource management Explicitly allocate resources amongst work Limit excessive or unexpected resource consumption Better request management Prioritize and track performance of work Tool: Optim Performance Manager Control all aspects of Workload Management through a single tool integrated into Data Studio and leverage advanced analysis in Performance Expert IBM Corporation

61 Workload Management Terminology Work Action Sets Identification and management of activities based on the type of work Service classes Single point of execution, monitoring, and control for work running on a data server Workload Identification of work to be assigned to a service class Thresholds Exception events that trigger an action for work on the data server Monitoring Light weight methods for determining workload status, trends and problem determination User Requests DB2 Execution Environment DB2 Engine Workload Service Class Workload Workload Service Class IBM Corporation

62 Example: DB2 WLM Monitoring Capabilities User/Application Requests Workload A Superclass DB2 Engine Subclass A Locations where Monitoring collection could be applied: Workload B Subclass B SQL using Table Functions Workload C Activity Information Default Workload Default User Service Class Aggregate Activity Info System Requests Default System Service Class IBM Corporation

63 WLM Dispatcher New in DB2 10: WLM Dispatcher is a CPU scheduling technology in the DB2 engine at instance level Allows for fine allocation of CPU amongst DB2 work executing in user and maintenance service classes It works in parallel with OS WLM mechanisms and supported for all DB2 LUW platforms Provides additional flexibility with use of hard and soft CPU shares, and CPU limits User Requests Data Server Instance Database Workload A1 30% Service Class A OS Other Processes User Requests User Requests User Requests Workload B1 Workload B2 Workload C1 50% 20% Service Class B Service Class C DB IBM Corporation 63

64 DB2 WLM vs. Oracle WLM DB2 DB2 WLM allows database priority to be based on the business criticality of the data Can control CPU via hard limits or soft limits (shares) Available for All OS Environments Can combine with AIX WLM and Linux WLM for enhanced CPU control ORACLE Has no mechanism to assign query resources based on data Database Resource Manager only provides control through soft CPU shares Has no integration with Operating System s WLM IBM Corporation

65 Summary High Availability and Disaster Recovery HADR Planned or unplanned outages DB2 purescale Continuous Availability Great Scalability DB2 Partitioning Improve performance for data warehousing DB2 Advanced Security LBAC & RCAC Protect your data with precise control DB2 purexml Natively work with XML Improve performance with Indexes, compression and inlining DB2 Workload Management Prioritize your workloads for maximal and minimal impact IBM Corporation

66 Questions? Subject: DB2 10 Workshop for Oracle Professionals Information Management 2012 IBM Corporation