The primary objective of a data protection strategy is to protect data frequently and consistently. The ever growing size of data, the diversity of systems that store that data and the high expectations of the creators and consumers of that data have made it almost impossible for a data protection strategy to achieve its primary objective. To make matters worse, organizations are demanding more from the data protection process than just backups and restores. For example, they want to repurpose the protected copies for other use cases such as analytics and reports.
Traditional backup solutions can’t meet all the demands of the organization and as a result, the market is flooded with solutions that promise higher availability, copy data management and rapid backup and recovery as well. The problem is that each of these is a separate solution, creating a second problem, data protection sprawl. The data center has more protection solutions than ever, yet it has less protection than ever.
A flash to flash to cloud primary storage solution can help reduce some of the pressure on the data protection process. First, because the first two tiers of the flash to flash to cloud design are flash based, it greatly reduces the need for multiple primary storage systems and in most cases eliminates them. The performance of the first two tiers of the system can support the large majority of workloads that the organization may have.
The first two tiers being flash also enables the design to rapidly protect and present data. The design counts on modern snapshot technology that redirects on write instead of copies on write. Storage systems that leverage redirected snapshots tend not to be limited in the number of snapshots they can take. Additionally, considering that all of these snapshots are, at least initially, on flash, the performance of the snapshotted volumes is extremely high. Thanks to flash-based snapshots, it is possible to protect data very quickly and repurpose it for a variety of tasks like testing, reporting and analytics.
Snapshots, by themselves, are not a complete data protection solution. If something goes wrong with the primary volume or the entire storage system then these protected copies are lost. It is necessary to replicate these snapshots to another system, which then applies a separate snapshot schedule against that data.
The secondary storage system does not need to be all-flash like the primary storage system. Instead of being a mix of high performance flash and high capacity flash, it could be flash and high capacity hard disk drives. As a third layer of protection, snapshotted volumes could be copied to the cloud / object storage tier for even more cost effective and longer lasting retention of data.
Providing some flash in the secondary system is a requirement though. Once all-flash is introduced for primary storage, users quickly become accustomed to the seemingly limitless performance of the all-flash system and applications are designed and scaled assuming they will always have flash performance at their disposal. If the secondary system is “no-flash” then users will complain about performance during the disaster and applications may not run. Using some flash in the form of a hybrid array will allow the most active applications and data to execute and minimize user complaints.
The use of a secondary system that is a mix of hard disk drives and flash drives is so cost effective that the organization may want to consider two—one on-premises and one at the DR site. The reality is that most outages are not the result of a complete data center failure. Having a second system on-premises, enables failover and recovered state to be on a local system, further minimizing user disruption.
The Role of Backup
A flash to flash to cloud primary storage design improves the front line of the data protection strategy significantly and cost effectively, but backup still has a role. The role of backup in the flash to flash to cloud architecture is to provide long-term retention of data, a copy of last resort and storage diversity.
The backup component of the architecture is able to create its backups from snapshot copies of data so it won’t impact production and it is designed to maintain copies for decades instead of years. Backup also becomes the copy of last resort, if something goes horribly wrong then the backup, while slow, may be a welcome data source. Finally, organizations should consider storage diversity. If the organization uses a backup system that supports tape drives, data can be copied to tape, again with no impact on production, enabling data to be stored on an entirely different type of media and one that is disconnected from networks.
Primary storage architectures have two objectives, meet the performance demands of the organization and meet its capacity requirements as well. Data protection is required to recover data with minimal loss and to do so rapidly as well as to retain information for as long as the organization requires. The organization also expects it to feed data to other processes such as test, reporting and analytics. The flash to flash to cloud architecture is able to meet all the demands and objectives of primary storage, as well as most of the demands of data protection, relieving pressure on backup and recovery software.