As a storage analyst, I am repeatedly asked to compare one storage technology solution to another. Several years ago, this wasn’t very difficult. In enterprise storage, there were really only a couple options: SAN or NAS. When comparing SAN storage, evaluations looked at performance, data resiliency features, and price. When comparing NAS solutions, technical comparisons likely emphasized price and capacity scaling, along with a myriad of different file system-specific features such as how granular the system could manage quotas or snapshots. Depending on how you look at it, those might be considered the good old days for a storage analyst. Comparing solutions was relatively easy.
Now, let’s fast forward to today. The storage technology landscape is not divided into SAN and NAS anymore. The emergence of unified storage about a decade ago consolidated protocols on storage systems, making the long established practice of dividing the market by protocol essentially moot. Also, there are just a lot more options now to consider. Object storage and protocols, such as S3 and Swift, have emerged to add another level of complexity. Solid-state or all flash storage adds another dimension, along with scale-out storage. This doesn’t even factor in the variety of deployment/consumption models for storage. Should storage functionality be delivered as software-defined storage, or public cloud, or part of a larger converged, or hyperconverged architecture? All these technologies present a double-edged sword for the IT infrastructure or storage administrator. IT teams now have more technologies and solutions to choose from, each providing benefits. At the same time, there are a lot more technologies to evaluate and understand when considering a solution.
Now that I have set the stage, let’s go back to the original question of how to compare these solutions. Some industry players, IT leaders, and even analysts have decided to isolate each category into a new market segment. Such as all-flash arrays are on one island, and scale-out storage systems are on another, with each category having a rigid qualification process that demands the solution meet a specific set of technology and architectural requirements to participate. Although those sorts of comparisons provide value, when making a buying decision, if you only evaluate the industry under that lens you are doing your business a disservice.
I wrote about this idea a while ago, but I think it makes sense to revisit this issue. At the end of the day, IT organizations are seeking a solution to store data for applications. And while different storage architectures can be more suited to serving certain application types than others, there is a tremendous amount of overlap. The obvious example is storage for a server virtualization environment. You could make an argument that under the right circumstances almost every storage technology type—SDS, hyperconverged, hybrid, NAS, SAN, all-flash, cloud, and even object—might make sense.
Another example is that of file storage. Looking to store unstructured data? A myriad of options are available. If you only follow some industry definitions, that means you need a scale-out shared-nothing hardware architecture, excluding dual controller offerings, a variety of SDS solutions, and some of the more recently emerged all-flash implementations for unstructured data. When doing a technical comparison it makes sense to draw the line somewhere, but for IT leaders looking to make a purchase, these rigid definitions are detrimental. What matters most is the value the end solution brings to the application and less the particulars about “how” the system is architected.
When I spoke on this topic previously I proposed a new spectrum for classifying data storage, focusing on solution benefits such as transactional performance or cost effective scalability, while also considering the deployment model. Is your IT organization seeking a more integrated deployment model, a converged model, a software-defined model, or a public cloud solution? And while we have seen a considerable amount of innovation between now and then, this model still holds up relatively well. (Though the emergence of cost effective all-flash storage for unstructured data may be breaking it a bit.)
So what is the take away? When evaluating storage for your next project or to replace an existing solution, do not limit the investigation to a rigid definition of technology and focus on the benefit that your organization is attempting to achieve. While I know this makes the job more complicated, it is the only way to ensure you and your business don’t miss out.