Not so long ago, buying storage was a simple matter of getting enough spinning disks of the right size to meet...
the performance and capacity needs of individual workloads.
In the past few years, the concept of a secondary storage structure has taken hold, in which data is assigned to primary storage or secondary storage depending on how critical it is to running workloads. Secondary storage contains all of the workloads that don't require primary storage to operate. Primary storage is then hyperfocused on speed and just enough capacity to run those critical workloads. Secondary storage runs everything else. In many cases, while secondary storage systems will likely still use flash-based SSDs to a degree, the focus for secondary storage is more around ensuring sufficient capacity for all of those secondary workloads.
In a world in which every workload is mission-critical to someone inside the organization, IT needs to be judicious in classifying these applications. In modern terms, secondary storage is, at its core, closely aligned with data protection. Data protection is, after all, generally considered the main purview of IT, which is charged with ensuring that the organization's data assets remain in production and available.
Taking a look at more traditional data protection mechanisms reveals that, in many cases, backup environments were a step-down from production ones. They didn't require fast flash drives, but they did need capacity. In other words, IT created a separate tier of infrastructure for data protection. The reason is that backup environments don't typically need the level of performance that is often demanded from mission-critical systems.
A similar process should be undertaken for all of the services and applications that IT intends to move to a secondary storage structure, and IT should determine if there needs to be additional tiers of infrastructure implemented to support differing needs For example, so-called tier-zero applications may require a tier zero-level infrastructure that consists of all-flash all the time. You may then have other applications that require more balance between performance and capacity but with a tilt toward performance or that require high CPU resources assigned. A separate infrastructure tier may be desirable rather than moving to converged secondary storage.
Common uses for a secondary storage structure
Let's take a look at the common use cases that may have previously required a separate tier of infrastructure but that can now be merged into a secondary storage structure.
File services. Don't underestimate the role of the file server. In some organizations, it's a nice, neat collection of well-organized files and folders; in others, it's a dumping ground for everything. Regardless, file services usually require a whole lot of capacity, even if they don't need a lot of performance. This performance pattern makes it an ideal use case for secondary storage.
Analytics. Data analytics requires copying data to a place where it can be analyzed. That place needs to have a lot of capacity. It may need speed as well, which makes it especially fortunate that current hyper-converged secondary storage systems include flash so that you don't need to deploy a completely separate infrastructure to support this use case.
An on-ramp to the multi-cloud universe. Of course, cloud itself is a tier of storage when you want to use it like that. But you can go a lot deeper as you bring converged secondary storage into the equation. With some converged secondary storage products, you can introduce to your organization a complete hybrid or multi-cloud fabric, enabling you to support a wider variety of use cases across a wider variety of environments. This capability often brings cloud-centric disaster recovery capabilities to your environment.
In many ways, a secondary storage structure is simply a consolidation of all of the nonprimary environments into a single tier that augments and runs alongside the primary environment. But, before you make the leap to secondary storage, take a look at what you're running in your organization, and consider the needs of those individual workloads.