Meeting Today s Enterprise Storage Challenges: A Better Backup and Recovery Solution
A Better Backup and Recovery Solution Challenges Facing Enterprise Storage Environments Today Technological progress can only go so far: the day still has only 24 hours, yet the demands that companies storage environments face keep growing. Shrinking backup windows, increased data volume, reduced staffing and higher productivity demands are only a few of the challenges that storage professionals face. The need to comply with new Sarbanes Oxley regulations in affected industries also raises the threshold of storage management complexity. Most shops have no time for data clean up and housekeeping, so that old data that is not being used is sitting around being backed up multiple times, increasing volumes even more. Most companies backing up to tape are hitting a wall with all these issues. Environments that had over 100 GB a day to back up five years ago are now facing up to a terabyte daily. Even the fastest new tape technologies will not be able to keep up with this level of growth. In an attempt to cope, many companies have turned to traditional incremental and differential backups on a daily basis. Even these are large and timeconsuming, and still require the regular performance of full backups, however frequently they are considered necessary, often weekly or monthly. These full backups can easily exceed a weekend backup window, and can also require onsite, full-time supervision. There is a better way. A Better Solution So what is the ideal data protection solution for today? One with faster backups, faster and more reliable restores, one that is efficient and leaves a small footprint on the system, as well as reducing the volume of secondary storage needed, and finally, one that scales easily. All in all, a solution that offers the best in business continuity, with minimum disruption in the case of a system failure. Business continuity in high-volume enterprises today requires frequent, fast backup and even faster recovery, of both data and applications. Traditional backup methods can t provide that. Two new business continuity objectives offer the measurement scale by which to judge a solution s success: A Recovery Point Objective (RPO) is defined as the optimal point in time of the recovered data, which would be as close as possible to the time of the moment of failure. The major obstacle to achieving the desired RPO is the inability to back up frequently without impacting other operations. A Recovery Time Objective (RTO) is defined as the optimum speed of recovery in event of a failure. The ultimate goal is recovery so fast that business is not affected. The major obstacle to reducing the RTO is the time needed to locate, assemble, and restore or remount backup data. The ideal is minimal RPO and RTO without costly mirroring solutions and storage expansion. The new technology which fulfills this ambitious goal combines the benefits of disk-based backup, continuous block-level incremental backup, synthesized virtual base and file-level restore to yield a next-generation data protection solution with revolutionary speed from both the backup and restore processes. Each component advantage will be further covered in detail, but in overview the entire new solution uses inexpensive ATA-based disk storage to provide fast, reliable access to backup data. Advanced backup technologies further exploit the advantages of diskbased storage. Block-level backup processing identifies changed data blocks at a level of granularity below file level and transfers only those blocks over the network to backup storage. Block-level backups transfer up to 90% less data than conventional file-level backup, significantly reducing the volume of bandwidth and storage needed. After an initial base backup, only blocks changed since the previous backup are transferred. Called incrementals forever, this technique eliminates the need for additional base backups. Instead, each instance of block-level incremental backup is automatically synthesized into a virtual base backup image. Therefore, a single restore job from a single backup instance is all that is required to restore a file, directory, or entire volume, including applications. Additionally, advanced technologies enable failover to the backup disk at the time of failure, thus allowing for near-instant recovery and continuous business operations without the need to transfer data. Data restore can then take place while business continues. One other important feature is database integration, which allows Page 2
for hot backup and recovery for SQL and Exchange databases. Why Back Up to Disk? Disk-to-disk backup offers many benefits over tape backup. It is faster, more easily scaleable and can be more reliable for restores than tape. In a traditional backup-to-tape scenario, there are SAN- and/or LANattached servers, which are backed up to a central tape library. Tapes are then sent offsite for disaster recovery purposes. As data volumes grow, however, scaling tape storage requires adding more tape drives, media and software licenses to meet the backup window, while adding disk space is a minor operation in comparison. A new approach is to insert an ATA-disk device that will be the primary backup target and then archive to tape for long-term storage. With primary backup available on the disk, the backup window to tape becomes immaterial; it can occur at any time without impacting the system. Disk is plentiful and inexpensive to add. It is also much faster for backup, and as data volumes increase, additional data streams can be generated to increase performance. And since the backup target is RAID, there is more builtin reliability in case of system failure. If a single disk fails, there are the others. This eliminates the potential issue of problem tape drives or media causing backup failures. In addition, speeds for restore are faster, since disk allows random access versus tape s sequential access. A further benefit is the elimination of the need to search an entire tape set to locate the files needed for restore. This approach can be expanded to add a second disk at an offsite location, furthering disaster recovery (DR) and business continuance plans. Data can be mirrored to the DR site instead of, or in addition to, offsiting tapes. Then a quick failover can be implemented in the case of a system outage. File-level vs. Blocklevel Incremental Backup There are two methods of accessing data on disk for incremental backup purposes: file-level and block-level. File-level occurs when the pieces of data on disk comprise a single entity (file) referenced by the host file system. In contrast, block-level is the smallest piece of data on disk that can be addressed. When files are created and/or modified the blocks on 950 900 850 150 100 50 0 BACKUP STORAGE GROWTH MODEL FILE-LEVEL BACKUP: WEEKLY BASE, DAILY INCREMENTAL BLOCK-LEVEL INCREMENTALS FOREVER MODEL ASSUMPTIONS: 12 TB STARTING 2% DAILY CHANGE 4% MONTHLY DATA GROWTH DAY 1 DAY 7 DAY 30 DAY 365 the disk associated with the file are changing. This is the lowest level of granularity associated with file changes. During file-level incremental backup, the backup agent communicates to the file system, which gets all the data from the disk and presents the file as a single entity. During block-level incremental backup, the backup agent communicates to a special driver, which gives it information below the file level about changes to a file s individual component blocks. Therefore, there are significant savings associated with block-level backups versus file-level backups. As illustrated in the two tables shown, an image-based blocklevel incremental significantly improves performance in a number of areas: reduced time, reduced resource usage, and reduced network bandwidth consumed. Since 907 TB 42 TB Figure 1 Storage Savings: Block-Level Incremental Versus Traditional Methods S T O R A G E S A V I N G S the data volume is smaller, it is faster to back up and takes up less storage space, and less network space getting there. This critically allows for more frequent backups and improved RPO. With the advanced technology that synthesizes each backup instance into a full backup, it is possible to quickly recover any file, directory, object or volume to any point-in-time with block- or file-level restore, or restore the entire system to any point-in-time. With the incrementals forever technology, all the performance improvements approach 90% or better. Page 3
100 90 80 70 60 50 40 30 20 10 0 Time (sec) CPU (%) File Backup Block-level Backup Figure 2 Block-Level Backup Reduces Elapsed Time, CPU Usage Backup Express Diskto-Disk Data Protection Solutions Backup Express Advanced Protection Manager is the advanced disk-to-disk data protection solution that exploits block-level incremental backup technology to provide the capabilities stated earlier in this paper. The Advanced Protection Manager is made up of five components, the first four of which work in conjunction with the Network Appliance NearStore SnapVault technology. Network Appliance (NetApp) SnapVault is a disk-based data protection solution that backs up filers and open systems (Windows, UNIX, Linux) to a secondary storage filer or NearStore system. Block-level incremental backups are stored on the SnapVault secondary as synthesized full backup images. Each component has its own specific advantage, and will be addressed in greater detail later. In overview now, however, they are: Syncsort Open System SnapVault (OSSV) Agent: Provides block-level incremental backup for high-speed backup of Windows systems with virtually no impact on other operations. It accesses the disk directly, bypassing the file system for extremely fast, efficient image-based backups to SnapVault. OSSV Database Backup: Provides block-level incremental backup of Exchange 2000/2003 and SQL Server 2000 databases to SnapVault. Uses Syncsort s OSSV agent for fast backup and restore. It displays SQL and Exchange volumes and databases in the Backup Express GUI for selectable backup and restore. Instant Availability: Enables rapid access to data backed up on SnapVault filers as read/write volumes. This allows near-instant recovery of critical Windows applications and data without the need to transfer data first. Express Image: Uses block-level technology for high-performance backup of Windows and Solaris systems to tape or storage-independent disk. This provides exceptional performance gains, especially for high-volume backups with many small files. Syncsort s OSSV Agent Syncsort OSSV is the Backup Express high-performance Open Systems SnapVault (OSSV) agent for Windows primary systems that supports backup from Windows NT/2000/XP/2003 primaries to SnapVault secondaries. Syncsort OSSV is the cornerstone of the new disk-to-disk management technologies, and enables all the other technological breakthroughs that Backup Express Advanced Protection Manager and NetApp s SnapVault solution provide. It is the Syncsort OSSV that exploits advanced disk-based backup technology to provide the block-level data transfer with filelevel restore, using incrementals forever. The key point with this process is that each backup instance is automatically created as a synthesized point-in-time base backup snapshot, eliminating the need to aggregate restores. This marks the first time in backup history that point-in-time restores are accomplished with a single restore instead of needing to restore the base, then the first incremental, then the second incremental, and so on. Syncsort OSSV image backups directly access changed blocks at the physical level, bypassing the file system. For large backups involving many files this method can dramatically reduce backup elapsed time and resource usage on the primary, compared to methods that interact with the file system. This makes it the fastest OSSV agent available. In addition, primary system snapshot capability for all supported Windows versions is included, so that additional software is not required for open file support. The Syncsort OSSV method also enables more frequent backups. Administrators can back up many times a day, without affecting users and other operations, compared to once or twice a day for other methods. This significantly improves the RPO, minimizing business disruption in the event of failure. The Backup Express browserbased GUI controls all Syncsort OSSV functions, as well as traditional backup functions. The GUI enables authorized administrators to configure and manage all primary and secondary nodes, define backup schedules, set retention policies, define backup jobs, browse and select for point-in-time restores, define and view reports, and more. All jobs, whether Sync- Page 4
sort OSSV or traditional, are tracked in a single catalog. Exchange/SQL OSSV Agents The Syncsort OSSV database agents provide high-performance, block-level incremental backups and fast, current, point-in-time restore of Exchange 2000/2003 and SQL Server 2000. The Syncsort OSSV database agents integrate with Microsoft APIs to insure data integrity and are aware of Exchange and SQL Server. Backup Express displays the databases as recognizable, selectable volumes in the backup and restore windows. All the advantages of Syncsort OSSV are applicable to database backup and restore, including physical disk-level access, blocklevel data movement and storage, direct restore from point-in-time snapshots, synthesized base backups from each backup instance, and the single catalog and efficient control from the Backup Express GUI. Syncsort OSSV allows restore granularity down to the database (SQL) or storage group (Exchange) level. More granular restores (Exchange item level) are possible as Backup Express provides for mounting data to Exchange. Individual mailbox and item restores can be done with Exchange utilities. Syncsort OSSV is a particularly valuable addition to enterprises with Exchange/SQL databases backed up on a NetApp filer using SnapManager, as it allows for non-netapp restore of Exchange/SQL. It also provides the flexibility of using Instant Availability in a failure situation, which snapshots alone cannot provide. Instant Availability Backup Express Instant Availability exploits iscsi target-attach technology to provide rapid temporary access from a Windows node to data stored on a SnapVault secondary. Instant Availability eliminates the need for data transfer in the event of data corruption or hardware failure by restoring access to data and applications almost instantly (under one minute), providing unparalled flexibility for disaster recovery scenarios. Instant Availability supports Windows 2000/XP/2003, including Exchange 2003 and SQL Server 2000. Once Instant Availability is activated, secondary storage volumes appear as local, fully accessible (read/write) drives. Files on the secondary are transparently available to users, who can continue to work while Syncsort OSSV recovery runs in parallel to restore data to primary storage. For quick recovery of individual files, users can easily copy from an Instant Availability drive to another drive on the local node. Instant Availability can also be used as an efficient backup verification tool because backup instances on the secondary system are readily usable from a standby node. In addition, Instant Availability is useful in development testing. By mapping the testing node to a backup copy of production data, the new application code can be tested against the read-only data. Administrators can use Instant Availability in conjunction with NetApp s SnapMirror for temporary access to data on mirrored SnapVault backups, without the need for data transfer. Experienced database administrators can also use Instant Availability and Syncsort OSSV in conjunction with database administrative procedures to provide point-in-time recovery without significant application downtime. ExpressDR ExpressDR, for Windows 2000/ XP/2003, provides high-performance bare-metal disaster recovery of both data and applications. ExpressDR uses routine Syncsort OSSV backups and eliminates the need for special disaster recovery backups. This has the major benefits of not requiring system interruption during the backup process, not requiring any system reboots, and not impacting users or system resources such as CPU, memory, temporary disk space, or network bandwidth. On the recovery side, ExpressDR eliminates the need to manually reinstall the operating system and root volume applications and streamlines the recovery process, significantly reducing the down time and optimizing RTO. ExpressDR can rapidly recover an entire or partial system. Depending on recovery needs, ExpressDR can restore the operating system, system settings, partition information, applications, and data all in a fraction of the time required by manual methods. In addition to fast recovery of a failed machine, experienced administrators can adapt ExpressDR to deploy an image to multiple machines and to continually update offsite secondary systems. ExpressDR is implemented in a simple three-step process that eliminates the need to install operating systems, apply patches, install applications, and apply configurations. Page 5
Integrated SnapVault Management Backup Express seamlessly integrates SnapVault within a total enterprise protection strategy. Consolidated control a single GUI, a single catalog, and a single backup server gives administrators the flexibility to design data protection environments that best leverage existing investments in tape and disk-based storage technologies. Network Appliance (NetApp) SnapVault is a diskbased data protection solution that backs up filers and open systems (Windows, UNIX, Linux) to a secondary storage filer or NearStore system. Block-level incremental backups are stored on the SnapVault secondary as synthesized full backup images. For backup of Windows primaries, Backup Express offers its own high-performance agent, Syncsort OSSV, described earlier. Backup Express Management allows implementation of powerful SnapVault technology diskbased backup and restore, blocklevel data transfer, incrementals forever, and synthesized base backup images without sacrificing manageability. Backup Express manages and tracks SnapVault hardware and software, OSSV agents and operations and traditional backup operations, all from the same master server and GUI. The intuitive Backup Express browser-based GUI enables authorized administrators to manage all data protection needs, whether SnapVault or traditional. This includes defining backups and backup schedules; setting retention policies; browsing and selecting for point-in-time restores at volume, directory, or file level; and configuring and managing all primary and secondary nodes. Unassisted restores can be performed by authorized users through the simple GUI restore screen. In addition, backup to tape can be performed for added protection. Backup Express NDMP backup jobs or NetApp s SnapMirror can be used to back up secondaries to removable media, if desired. Express Image Express Image is the only component of Advanced Protection Manager that does not work exclusively with NetApp technology. Express Image provides high-performance, block-level incremental backup to tape or storage-independent disk, as well as volume level backup with volume, directory, or file restore. It supports Windows NT/2000/XP/2003 and Solaris 2.8 and higher with UFS. During Express Image backup, online applications continue to use the file system, unaffected by backup processing. Backup performs at raw-disk speed, even for volumes containing many small files. For tape backup, or diskbased backup where SnapVault is not available, Express Image provides the same block-level backup techniques and direct, physicallevel disk access. A combination of techniques provides speed and resource efficiency: only allocated blocks are backed up; data movement is at block-level, not file level; and physical-level disk access bypasses the file system for significant increase in performance. The result is faster backup and reduced consumption of server resources such as CPU, memory, and temporary disk space. Express Image can reduce CPU usage by more than 90%, and the performance benefits increase with the size and the number of files being backed up. For base backup involving large number of files, Express Image is orders of magnitude faster than file-level backup. Express Image provides a flexible, high-performance alternative to traditional backups. Base, incremental, and differential backups all use block-level backup with filelevel restore while live, open-file image backup does not interfere with applications or file systems. Backups can be directed to local and/or remote tape, disk, or filer and are controlled and managed through the same Backup Express GUI and the single Backup Express catalog. For rapid recovery in the event of a disaster, physical (block-level) restores can be done at the volume level, while logical (file-level) restores can be done at volume, directory, or file level and support such extended attributes as ACLs. Conclusion This has been but a short overview of the benefits and advantages that disk-based incremental backup can bring to almost any storage environment. Whether used alone or in conjunction with more traditional methods, this next-generation technology significantly speeds and simplifies the backup and restore process. The combined benefits of disk-to-disk backup, continuous block-level incremental backup and synthesized base backups yield the drastic reductions in RPO and RTO that are the benchmark of success. The ability to do more frequent backups, without impacting other operations, significantly decreases the window in which data can be lost from a day to an hour in many situations. This near instantaneous RPO drastically decreases the disruption that any Page 6
single failure will cause an enterprise. Similarly, the ease and speed of restores reduces the RTO to minutes even under a minute when Instant Availability is used which can truly be considered a recovery timeframe so fast that business is not affected at all. Seamless business continuity is ensured making this the data protection choice for the future. Page 7
50 Tice Boulevard Woodcliff Lake, NJ 07677 www.syncsort.com 201-930-8200 Syncsort Incorporated, 2005 All rights reserved. Backup Express is a trademark of Syncsort Incorporated. SnapVault and NetApp and NearStore and the Network Appliance logo are registered trademarks and Network Appliance is a trademark of Network Appliance, Inc. in the U.S. and other countries. All other company and product names used herein may be the trademarks of their respective companies.