Last week, I, along with my colleagues Mark Peters and Tony Palmer, were given the opportunity to attend the Pure//Accelerate user conference in sunny San Francisco. You can check out Mr. Peters thoughts and our on location video here, but I would like to add my own perspective on the event and Pure’s announcements.
I wanted to take a moment to reiterate just how large this event was relative to the size of Pure’s existing customer base. Estimates that attendance may have represented half of Pure’s existing customer may actually be on the conservative side. While it may be easy to get caught up in the passionate fan base that Pure Storage has created, or the strong net promoter score (79) that the company has maintained despite their growth, the storage industry has seen this before. Over the years, the storage industry has seen a number of smaller storage players with a strong product and a passionate fan base. Pure Storage never would have gotten to where they are today if it weren’t for those things, but the more important question for me to understand at the event was what is next for the company.
We got a glimpse of what's next when Pure Storage unveiled its two new products. The first of which, the FlashArray//M10, is what you would expect from an entry-level all-flash array, continuing Pure’s strategy to make flash storage more affordable and more available. The second announcement, the FlashBlade, is far more interesting. Not only in what the product can do, but in what it means for the company and the industry moving forward. A key point in the evolution of any company is the transition from a successful product to a successful product portfolio. This step can be — and often is — the defining moment in a company’s trajectory.
But before we get to my thoughts on FlashBlade, let me concede my pre-existing biases on flash storage and the all–flash data center concept. For flash storage, based on the research we have done at ESG as well as my own direct conversations with IT leaders, I am a strong believer in the dramatic benefits offered by the technology. If you are not using flash storage in some fashion, you are very likely doing yourself, your IT organization, and your business a disservice.
The all-flash data center is a slightly different story. I fully expect that someday we may look back on the use of mechanical spinning media as an odd blip over the course of the history, like the 8-track or the vacuum tube. Over the near future, however, I believe(d) that the wide variation needed in IT workloads, with different requirements in terms of performance, capacity, and resiliency, will demand the ability to leverage different storage media types. Designing an efficient data center, therefore, would require the ability to utilize the right media option for right workloads; flash storage for transaction performance and spinning high capacity storage for cold unstructured large content storage repositories.
Pure Storage’s FlashBlade, however, is changing my mind. The all–flash data center may be much closer than I had previously thought. While our technology experts at ESG have not yet had a chance to get their hands on a FlashBlade system, the claims are substantial. Pure Storage is claiming up to 1.6PB of effective all-flash storage capacity and the ability to store 10s of billions of files/objects in a 4U form-factor that draws about the same amount power as a microwave oven and costs less than $1 per usable GB of capacity. While the price point and the use of S3 and NFS protocols may lead some to see the FlashBlade as Pure’s expansion into secondary cold storage workloads such as archive or back-up, the all-flash performance offers the potential to do far more. More contemporary workloads, such as private cloud, big data analytics, and the Internet of Things all often demand large content repositories with performance demands far higher than those of the traditional long term cold archive. FlashBlade may still likely be overkill for some unstructured workloads, but the price point combined with the savings in density and power could make a compelling case even for colder data archives.
It may still be some time before we get to the fully all-flash data center, but Pure’s FlashBlade may have just cut five or ten years off the journey.