Cracking the Oracle RMAN Code Delivers 4-5X Improvements in Deduplication Ratios
IDC recently announced in its report, "The Diverse and Exploding Digital Universe: An Updated Forecast of Worldwide Information Growth Through 2011", that the digital universe is bigger and growing more rapidly than originally estimated. The research showed an annual growth rate of 60% and stated that the digital universe will grow to a projected 1.8 zettabytes by 2011. While a lot of this data is in an unstructured format, you can rest assured more than a fair share of it will end up in corporate databases. With this explosive rate of data growth, many database and storage administrators will have their hands full managing backup and recovery windows.
Since Oracle has a commanding lead in the database market, it only makes sense to talk about how deduplication can help in an Oracle environment. From a historical perspective, it's interesting to step back and see just how Oracle backups have progressed from exports or cold backups, advancing slowly into hot backups and eventually to the adoption of RMAN (Recovery Manager). Each of these stages was often met with criticism, caution, and a bit of confusion. Regardless, Oracle's RMAN utility has been around quite a while and has become the near de facto standard for initiating backups, restores, and recoveries of an Oracle database.
The amount of data in databases has long been a target of data reduction techniques, most notably compression--and some database files compress very well. But applying data deduplication to database files has been more problematic with most dedupe systems not providing much additional reduction beyond compression (which is usually part of the dedupe process).
The issue with deduplication is that the repeated patterns that deduplication finds can be scrambled or obscured in a dynamic database file by metadata (section markers, end of file tags, etc.) that backup applications insert when they process the backup. If a deduplication system can differentiate this metadata from the underlying content, deduplication rates can be increased dramatically.
The folks at Quantum recently announced that they had cracked the code for Oracle RMAN, allowing their DXi deduplication appliances to screen out the metadata in RMAN files and greatly increase their dedupe results - early installations are showing 4 to 5X increases - although the rate will obviously vary based on the amount of changed data between backups. Quantum tells us that the efficiency works for RMAN when it is being used as a primary application, and it has verified similar results when RMAN is used as part of the backup in several other backup applications, including Symantec NetBackup, Oracle Secure Backup, and CommVault products when CommVault Simpana is used in its Single-Instance mode.
Deduplication is emerging as a critical capability for customers who need to effectively store, manage and protect databases. The Quantum initiative to crack applications that backup our Oracle databases is highly needed to reduce the amount of data that organizations need to store and free up network traffic between source and remote sites while providing a complete solution that matches the backup needs of Oracle within the data center.
Since Oracle has a commanding lead in the database market, it only makes sense to talk about how deduplication can help in an Oracle environment. From a historical perspective, it's interesting to step back and see just how Oracle backups have progressed from exports or cold backups, advancing slowly into hot backups and eventually to the adoption of RMAN (Recovery Manager). Each of these stages was often met with criticism, caution, and a bit of confusion. Regardless, Oracle's RMAN utility has been around quite a while and has become the near de facto standard for initiating backups, restores, and recoveries of an Oracle database.
The amount of data in databases has long been a target of data reduction techniques, most notably compression--and some database files compress very well. But applying data deduplication to database files has been more problematic with most dedupe systems not providing much additional reduction beyond compression (which is usually part of the dedupe process).
The issue with deduplication is that the repeated patterns that deduplication finds can be scrambled or obscured in a dynamic database file by metadata (section markers, end of file tags, etc.) that backup applications insert when they process the backup. If a deduplication system can differentiate this metadata from the underlying content, deduplication rates can be increased dramatically.
The folks at Quantum recently announced that they had cracked the code for Oracle RMAN, allowing their DXi deduplication appliances to screen out the metadata in RMAN files and greatly increase their dedupe results - early installations are showing 4 to 5X increases - although the rate will obviously vary based on the amount of changed data between backups. Quantum tells us that the efficiency works for RMAN when it is being used as a primary application, and it has verified similar results when RMAN is used as part of the backup in several other backup applications, including Symantec NetBackup, Oracle Secure Backup, and CommVault products when CommVault Simpana is used in its Single-Instance mode.
Deduplication is emerging as a critical capability for customers who need to effectively store, manage and protect databases. The Quantum initiative to crack applications that backup our Oracle databases is highly needed to reduce the amount of data that organizations need to store and free up network traffic between source and remote sites while providing a complete solution that matches the backup needs of Oracle within the data center.
Leave a comment