Before all-flash hybrid arrays existed, systems made up of flash, and high performance hard disk drives were the go-to option for organizations looking to solve performance problems. However, hybrid systems’ dominance was short lived as all-flash arrays came to market and their prices continued to drop, all while obliterating most performance problems. Nevertheless, hybrid arrays still have a role to play; they may be the ultimate secondary storage solution.
The All-Flash Problems with Old Data and Disaster Recovery
While all-flash arrays have significantly come down in price, these systems are still at a premium compared to hybrid systems. The price delta creates problems for data centers looking to go all-flash. First, what about old data? Does the data center want to store data users may never access again on flash storage? It just seems wrong to have dormant data sitting idle on flash storage.
The second problem for the data center is all-flash disaster recovery. While most all-flash systems now have built-in off-site replication capabilities, almost all of them require replication to another system from the same vendor. If the vendor only sells all-flash arrays, that means that the second system, which will more than likely sit unused at the DR site, is another expensive all-flash array.
The Problem with Secondary Storage
Secondary storage solutions are now flooding the market. These systems are designed to be lower cost and more scalable than primary storage solutions and promise to be the storage system for a variety of use cases like archive, backup and user home directories. IT Planners can implement a variety of third-party solutions that will replicate, archive or backup data to these systems.
The problem is that these systems in the all-flash era have one big problem, lack of production performance that is similar to all-flash. While a production all-flash system is running users get used to all-flash performance, and developers create applications that expect flash performance.
If the IT planners implement a third-party software replication program so they can replicate from an all-flash array to a lower cost hard disk-based array, users may complain about slower performance during the disaster and applications expecting flash performance simply may not run. The lack of DR system performance forces the organization to buy a new all-flash array and migrate data to it when resources are already stretched thin because of the disaster.
The Hybrid Solution
A better alternative may be to use a hybrid (flash and HDD) array from the same vendor as the all-flash array, in the disaster recovery site. Because they are from the same vendor, they are more likely to be able to replicate to each other. The hybrid system, since it is at the DR site, can have a relatively small flash tier and a very large HDD tier. IT can store old data on it, and also replicate active data to the system. Because of its lack of activity, the data will be stored mostly on the hard disk drives.
The flash tier though becomes very useful in the event of a disaster. When a disaster is declared, the secondary storage system becomes a primary storage system and needs to run production applications at near production performance. The flash tier makes it possible to deliver near all-flash performance during the disaster. If it looks like the disaster will take some time to resolve, the organization can add additional flash capacity, upgrading performance without having to migrate data, at one of the worst times possible.
The hybrid system, because it can leverage high capacity hard disk drives, can meet the scaling requirements of most organizations. Some organizations might look into adding an object storage system to their storage architecture, as well, for the long term storage of data.
For some organizations, hybrid storage may be past its prime. However, this is only for the production use case. Using a hybrid storage system as a secondary storage target to a production all-flash array reduces costs and improves the organization’s ability to recover from a failure, and to sustain expected performance during a disaster.