Scanrail - Fotolia
The answer to the question of whether flash storage should be used for backups varies greatly based on who you ask. Your storage vendor is almost certain to recommend that you replace your existing backup media with a fancy new all-flash array. Conversely, your organization's CFO is more likely to tell you that you can get at least another year out of your existing backup product. The real answer to the question of the need for flash backup often lies somewhere in between.
Capacity has long been the most important backup storage factor. In the case of disk-based backups, the storage array needs to accommodate the current working set of data and multiple recovery points. Data stored on the disks varies depending on a number of factors, including the amount of data the organization needs to protect and the overall backup architecture. In some cases, a disk-based data backup may contain a full copy of the organization's data plus several recovery points; in others, only the most current data resides on disk-based backup.
Certainly, capacity requirements figure into the decision of whether it is better to use flash backup or spinning media storage for backups. After all, capacity requirements determine the overall cost. Granted, data reduction methods can bring down the cost per gigabyte of storage, but ultimately, it is the capacity requirements that drive cost. Even so, the decision to adopt flash storage is commonly based on performance requirements.
There is little doubt that replacing your backup storage array's spinning media with a flash-based storage backup option will improve performance. The question is: What level of performance does the backup storage array really need? If an organization is operating a legacy continuous data protection (CDP) backup, an all-flash array may be overkill because the available network bandwidth for data coming into the backup server can restrict backup performance to a level that is far below what the flash array is capable of delivering.
In these types of situations, it may be better to take advantage of tiered storage rather than adopting an all-flash array. Tiered storage can be configured to use flash storage as a read/write buffer. When properly implemented, this technique can deliver flash-like performance without the cost of implementing flash backup based on an all-flash array.
Another important factor to consider is that modern backup offerings provide capabilities that go far beyond those of legacy CDP. Many backup products, for instance, include instant recovery capabilities that allow virtual machines (VMs) to be mounted and run directly from backup storage while a recovery happens in the background.
Instant recovery greatly increases the storage IOPS requirement for a backup storage array. This is especially true if instant recovery is performed on multiple VMs simultaneously. If you need to perform multiple, concurrent instant recoveries and your existing backup storage can't accommodate the required IOPS, you may benefit from upgrading to flash backup via an all-flash array.
Flash backup is one reason flash supplants hybrid
Is all-flash storage a better choice than hybrid?
A look at the startups in the all-flash storage market
Explore the best flash backup use cases
Dig Deeper on All-flash arrays
Related Q&A from Brien Posey
It's critical for an organization to know what data it needs to retain and where to store it. Some data is required for retention by law, so a ... Continue Reading
Decentralized storage technology can be confusing and complicated. These best practices, however, can help with implementation in enterprise IT ... Continue Reading
Organizational resilience encompasses everything a company needs to run in times of crisis. These examples show how businesses handle tough ... Continue Reading