Database applications are typically the lifeblood of organizations, supporting revenue generating activities and/or mission critical business data analysis. Database administrators are the stewards of this infrastructure and often have an almost personal interest in the welfare of these precious information assets. For this reason, many DBAs have historically performed their own backups and restore operations, running outside of the corporate data protection process. This has created silos of backup storage and often a duplication of infrastructure.
With the popularity of server virtualization and handheld compute devices, plus a more mobile workforce, IT organizations can find themselves running more data protection solutions than in the past. At the same time, more application owners (like DBAs) want control over their own data and many traditional backup applications seem unable to provide a solution that will handle everything.
The result is often multiple backup processes to protect the variety of applications and endpoints in play, plus more redundancy in the data protection process. But IT needs more efficiency and consolidation, not less. They need to meet the needs of their internal customers, like DBAs, who want control in order to manage the most critical information assets, but do it economically. EMC now provides that capability with Data Domain Boost software.
What is Data Domain Boost?
DD Boost is a software option that enables Data Domain systems to distribute parts of the deduplication process to supported client/servers to improve speed and efficiency. It shifts parts of the dedupe process to the application client or backup server and enables it to significantly reduce the size of the data set actually sent to the Data Domain system. This results in less data being sent across the network and less overhead on the client CPU, since the dedupe process is less CPU intensive than the data transfer process. The overall result is a reduction in the back up window and improved efficiency in handling and storing backup and replication data.
Deduplication moving Upstream
When first introduced, DD Boost moved part of the deduplication process into the backup server like EMC’s NetWorker, which provided the efficiency gains of reducing data before it’s sent to the Data Domain appliance. But Oracle was still sending its full Oracle Recovery Manager (RMAN) backups over the LAN to the backup server, transmitting these larger data sets and consuming more network resources. (See figure below)
DD Boost has now been moved further upstream, and integrated with RMAN, providing data reduction and efficiency at the Oracle server. Using the DD Boost plug-in now resident on the Oracle server, RMAN can run initial data segmentation and comparison checks to the Data Domain system to determine, locally which data is redundant. Then it runs the fingerprinting and compression processes on these segments before sending them to the Data Domain system.
With DD Boost integration RMAN can now send these pre-deduplicated backups directly to the Data Domain appliance. The DBA now no longer has to be dependent on the backup administrator to protect mission critical databases. And, although DD Boost runs on the client Oracle server it can still leverage the deduplication catalog for all backups stored on the Data Domain appliance, without involving the backup administrator. (See figure below)
DD Boost can improve backup performance of Oracle RMAN by up to 50%, reducing network bandwidth consumption by up to 99% and alleviating up to 40% of the CPU load on the Oracle server during the backup process. This results in an aggregate throughput of over 25TB per hour for Oracle backups and a 10x to 30x reduction in the amount of backup storage capacity consumed.
In addition to improving backup performance and efficiency, DD Boost gives Oracle DBAs complete control over their own data protection; backup, recovery and DR. They can run database backups independent of the backup application in place for the organization at large, creating a separate backup data store and managing their own restores.
When a remote Data Domain appliance has been implemented to provide for disaster recovery, DD Boost ensures full catalog consistency from RMAN. And, it enables the Oracle administrator to manage backup copies. Not only can on-site copies be managed by RMAN they can manage the entire replication process between two Data Domain systems, meaning that both on-site and off-site data protection is managed via RMAN.
With DD Boost, the organization can still leverage the Data Domain appliance as a central backup repository across the enterprise and eliminate the expense of ‘siloed’ Oracle backups. Deduplication is managed across all data sets, both RMAN backups and those taken by the main backup application, in order to maximize data reduction ratios.
Storage Switzerland Take
Dedicated Oracle backup is a fact of life for many organizations. It gives DBAs the control they want over the protection of their databases, but at the cost of running two backup processes. With the advent of server virtualization, mobile computing, bring-your-own devices, etc, this situation is poised to get worse.
DD Boost gave Data Domain users the advantages of client-side deduplication, improving efficiency as it reduced data handled by the client server and sent across the network. Putting DD Boost into the Oracle server moves parts of that dedupe process further upstream, giving RMAN users a chance to benefit from this technology as well. Now Oracle DBAs can reduce the amount of data that RMAN has to send into the backup stream and the amount of data that the network has to handle.
It can also allow RMAN to back up pre-deduplicated data directly to the Data Domain appliance, bypassing the backup server altogether. The result is a win-win situation. Oracle DBAs retain their control over backups and more importantly replication. As a result the backup and disaster recovery infrastructure is simplified and the entire backup process handles less data.
EMC is a client of Storage Switzerland