A lot of people, especially financial analysts and investors, have been missing the point for far too long with respect to enterprise flash storage vendors. When the flash revolution started, many thought that somehow simply replacing disk drives with flash memory modules was going to be the end-all of enterprise storage. That’s not how it works. Sustainable success in enterprise storage has never come from the components that go into a product, it has come from the way those components are used.
The fact is, flash memory modules have minimal value to enterprise customers, otherwise they would be buying a lot more of them and building their own flash storage systems. There is simply no way to make a DIY flash array that uses flash memory as effectively as arrays designed by vendors who have to compete with each other. You could argue that there have been some successful DIY storage made with Software Defined Storage (SDS) products running on servers, but realistically, SDS is neither DIY, nor is it as capable as vendor-engineered flash arrays.
Flash memory modules provide new component-level performance capabilities and new price points as well as introducing new behaviors in how components wear out. There is enormous potential left to figure out how to optimize the positive and minimize the negative aspects of flash memory. The earliest movers in the enterprise flash business were Nimble Storage and Pure Storage, both choosing different approaches to optimize and minimize flash’s pros and cons. Many other companies followed, including Tegile, SolidFire, Tintri and Kaminario, as well as EMC, IBM, HPE and Hitachi – each with different approaches to achieve the best flash recipes.
Implementing flash memory in enterprise storage arrays was never going to be a sustainable advantage. It was an inflection point that temporarily removed the barriers to entry for new vendors. In some cases flash created a disadvantage for legacy disk-optimized array architectures, but those will be overcome in time. It’s a bit cruel to discover that the technology a company built and refined for so many years suddenly turned into a limitation, but leapfrog technology has a way of doing that. For example, an architecture that stripes data across a large number of disk drives to boost transaction performance can introduce internal hop latencies that are unrecognized in disk drive-based systems but may become noticeable in a flash-based system.
With flash re-leveling the playing field in the enterprise storage industry, the challenge for all vendors is finding sustainable technology advantages beyond flash performance. Vendors are busy trying to establish distance with what they believe are their advantages while they simultaneously try to close gaps with what they view as competitive weaknesses. Investors tend to like simple equations such as flash=good, disk=bad. Discussions about using inline dedupe versus post-processing and mixing and matching All-Flash arrays with hybrid arrays for disaster recovery protection are too detailed for investors to appreciate but these are the sorts of things that matter most to customers who are simultaneously trying to steer through the changes in compute infrastructures with virtualization, containers, unikernels and the cloud.
When vendors talk about longer sales cycles affecting their financial results, it’s because customers in many cases are taking more time to figure out which vendor has the best solution for their needs. I can’t recall seeing as much uncertainty in sales scenarios in our industry as we have today. Customers are struggling to make up their minds and vendors who come into a deal late in the process have a much better chance of winning it than they used to. In other words, the sales environment for enterprise storage has become far less predictable than it was a year ago. In that sense, it is not easy being a public storage company right now because market visibility is murky at best.
The entire IT industry, including the enterprise storage business is in turmoil. Flash storage is growing rapidly, but unevenly. This situation has been misinterpreted by Wall Street and investors will likely continue dole out unwarranted excitement and unfair punishment. Flash memory created a new baseline for storage performance but the technology that unlocks the value of that performance potential is far subtler and still under furious development.