Q&A: Pure Storage CEO Discusses Flash Future, Competitors, And Vision To Be A Data Center Power

Pure Storage And The Future Of Flash

All-flash storage vendor Pure Storage celebrated its status as the newest publicly listed storage vendor by holding its first customer and channel event, Pure//Accelerate 2016.

The event, held in March in the shadow of San Francisco's AT&T Park, gave the company the opportunity the opportunity to not only show off its new FlashBlade all-flash array for unstructured data, but to spend time with channel partners and customers talking about the company's vision.

Pure Storage CEO Scott Dietzen sat down with CRN for a frank discussion of the new technology, the future of the flash storage industry and the company's place in that future, and the competition, especially NetApp's acquisition of SolidFire.

Here's a look at where Pure Storage fits in the data center today and where it is heading in the future.

What's the big emphasis at this first-ever Pure//Accelerate conference?

We're talking about innovations that Pure has been working on, in some cases, for three years. The overall theme of the conference is putting the tech to work in changing our customers' business. We're featuring customer case studies wherever we can. In fact, the Mercedes Formula One team that won first and second place last year was just talking about how they used Pure Storage in their Internet of Things. They actually collect analytics off of the cars in order to understand how the cars are performing. And then they run very extensive algorithms over that data they collect. And it's got to be really fast, because they're adjusting the cars in real time while they're racing. It was a great illustration of the proliferation of data and the need for very high-performance access to large data sets.

What does Pure mean to the data center? And what does the data center mean to Pure?

Roughly half of the dollars that are spent on storage is on the performance-optimized systems that support things like databases and virtual machines. We've had phenomenal success in that transition to the point that all the incumbents are saying all performance data belongs on flash, not on mechanical disk systems.

Today, it's about the rest of the data in the data center. Historically, there's been a lot of data stored on capacity-oriented systems with slower disk, often SATA disks, or 7,200-rpm disk. We're now disrupting that market by delivering an all-flash solution that offers dramatically better performance, better density, lower total cost of ownership. ...

I would say a key part of the discussion today is how do we unleash the value of the rest of the data that's still locked up on mechanical disk.

A few years ago, some analyst said the world can't possibly produce enough flash to make enough SSDs to meet demand.

And there was also a discussion [that] flash will always be more expensive than mechanical disk. There's no way that flash can ever compete. And yet ...

But now you're talking about under $1 per gigabyte.

When we got started, the consumer-grade flash we used was four times more expensive than the fastest 15-K [rpm] disk per-gigabyte. We worked very hard on data reduction in order to be able to shrink that cost. Did a bunch of other things as well.

Cutting The Cost Of Flash Storage (continued)

Flash has dropped roughly four times faster in cost than fast disk. Without any total cost of ownership advantage, without any data reduction, flash today is cheaper raw than a fast mechanical 15-K rpm disk. On top of that, we average five times data reduction and 10X more efficient in power and cooling. ...

We don't have to compete on total cost of ownership to make the acquisition price well below that of fast disk. The same thing is going to happen in this market for capacity-oriented storage.

You can't beat Moore's Law. It keeps right on progressing. Storage, in a mechanical world, wasn't on Moore's Law. That's ultimately what's driving this discussion. We saw [this] in advance, and have been crafting the software and the hardware architecture necessary to deliver on that promise of flash.

Yet today, you are also dealing in a very competitive environment with not only legacy vendors who have all committed to all-flash portfolios, but also to a number of startups like yours. How is that environment shaping up for Pure?

We are thrilled. I mean, the CEO of EMC's storage business and the CEO of NetApp both said earlier this year all primary data should go on flash. That was the first time the two vendors with the most to lose in this transition from mechanical to solid state storage made that announcement publicly.

We are seeing this shift accelerate. In our view, the whole of the storage industry is basically coming to the playing field that we laid out back in 2011, and that we just re-defined here today. And we feel we are so well positioned to deliver customers more value. Their 20-year-old designs, if you just stuff them full of SSDs, you can't get the full value of flash. Each of those flash drives does 64 things in parallel, and it does it all in microseconds. The software for mechanical disk is designed to do one thing at a time.

Dealing With Legacy Competition (Continued)

None of the legacy solutions are offering anything close to the effective data reduction. They're not offering effective copy protection because they're still doing real physical copying as opposed to using the same deduplication underpinnings to make it vastly more efficient. They can't support the modern protocols like NVMe, which is an addressing scheme that is better suited to flash than mechanical disk.

And, by the way, that's reflected in our customer satisfaction scores. We have an audited Net Promoter score from Satmetrix, which does the surveys for many of the top companies of 79, which is 63 points higher, on a 100-point scale, than the industry average.

If you look at the competitive environment, you'll see IBM with its TMS [Texas Memory Systems], EMC with XtremIO and DSSD, both claiming to be the sales leaders. Legacy vendors must be doing something right.

So keep in mind the legacy vendors have really large footprints. They have tens of thousands of customers around the world, and not all of those customers have yet heard of Pure Storage. What we have is a highly differentiated product. Customers try out Pure Storage, they love it, and they buy more. At the top level of our customers, they spend an initial dollar, they will spend 12 more dollars in the next 18 months. We're not seeing that kind of uptick for any of the competing technologies. The competing technologies don't have the simplicity, the resiliency that's essential for cloud-capable data center deployment. We just have to get customers to try the product. ... We're right on track to be building the No. 1 brand in storage, but it's going to take us a few years.

However, Pure Storage can be characterized as a one-trick pony. You have a couple different types of flash arrays, but still your whole business is built around one technology. Isn't that a risk for a fast-growing business?

Pure was a single-product company until [this month]. We announced our second product, which is also a flash storage device. But it's new software, new hardware, a whole different design. And it's competing in the market for file, scale-out file, and object storage as opposed to the block storage market where our primary product competes.

With those two products, we're in a position to cover the storage market, almost in its entirety, if you leave out for now archive and backup data. ... That just leaves an unbelievable growth opportunity in front of us.

When I look at the incumbents, I'd much rather be in our position. When you take Dell and combine it with EMC, they will have nine different all-flash storage offerings.

Assuming no consolidation of offerings ...

I would probably argue there's six too many. And if we're going to see some dramatic consolidation, that's still to come. Investors and customers are all concerned about how do I pick the winners from the losers? Which one's going to work? We think this is an environment where Pure has become one of if not the safest bet in the whole storage industry.

When you say nine different platforms between EMC and Dell, you're including all-flash versions of legacy storage like Compellent?

Sure. All-flash VMAX. All-flash VNX. XtremIO. DSSD. We've also got all-flash VSAN in the market. Compellent and EqualLogic also have all-flash versions. All-flash Isilon. That's not comprehensive, but it gives you a sense.

And when you use which technology for which problem is an exercise being left to the sales force and to the customers. But it's a very confusing situation. We think the clarity of our model provides a much better fit.

One of your competitors, SolidFire, gave a similar message: single platform addressing a wide range of the market. And now they were acquired by NetApp. What is Pure doing different?

The FlashBlade solution we launched [this month] is a much denser scale-out architecture. It provides a much higher level of performance. It allows you to pack more than a petabyte-and-a-half of data into a single 4U chassis. That's unlike anything we're seeing from any of the other vendors. Hardware efficiency matters hugely. It's how we've managed to build a business. Our business, we're in the high-60s in terms of margins. And we're doing that even though we can underprice most of the competing products.

If you compare our current FlashArray products to, for example, the corresponding technologies at EMC or NetApp or so on, they will typically require two to two-and-a-half times the flash and four to five times the CPU, four to five times the DRAM, 20 times the interconnects. ... This proliferation of hardware makes them so much less efficient, and really challenges their business to make them cost-effective.

Doesn't NetApp's acquisition of SolidFire give it something in a similar vein?

No. SolidFire is similarly inefficient in its use of hardware. They do things like requiring mirroring to protect the flash. That's a very simple thing. But if you're going to keep two copies, which is the minimum we believe [necessary] for reliability for an enterprise workload, we see two and a half times the flash. If you look at a half-petabyte configuration of SolidFire, you're talking more than two-dozen nodes. And that leads to a dramatic amount of hardware that has to be provisioned, and that just prices them out of the market. ...

Obviously, they [now] have a much bigger company and bigger channel behind them. The position that they play in is somewhere in between FlashArray and FlashBlade. And we think these products are much, much better than what they do at SolidFire.

So how has Pure Storage managed to stay independent and not get acquired?

We've had overtures. We've refused to have this conversation in all cases. We believe we're on track to build the No. 1 brand in storage worldwide. We are most likely to be successful as an independent. As an independent, we can offer not just dramatically better product, but we can offer dramatically better customer service than the competition. ...

It's such a highly differentiated product. We have a situation where most of our competitors are getting failing grades. We have higher customer satisfaction than an Apple iPhone. For a storage product, that's a pretty surprising thing. People love their iPhones. But they love their Pure Storage even more.

Let's assume for now that Pure Storage has no plans to get acquired. ...

We have north of $600 million in the bank. The business is cash-flow-positive to the tune of $30 million in the fourth quarter. This is the fastest-growing storage company in history. The recipe is all set for us to continue this great run we've had as an independent.

Let's turn the question around then. Is there anything out there that Pure Storage would like to acquire to create more opportunities for its customers and partners?

Certainly in our view, we were several times larger than SolidFire, and we were growing much faster. And that's part of the reason that they entertained merging with another storage player, because the IPO market is not so friendly to companies that don't have the growth and the differentiation.

I'm not going to point at any specific companies, but I do think there's always interesting innovations going on in the market. We're going to work very hard to not be disrupted. I think we're in a position that, if there are attractive entities out there, that we're going to get first looks at those companies. The thing that's so great about Pure is, there's so much growth ahead in our software. Our business is fundamentally healthy and growing, whereas most of our competitors' businesses are actually shrinking.

Then let me ask you this way: What types of companies would you want to call you and talk with you about their innovations?

These are the kind of things that we don't talk about publicly. We never want to tip our hands about how we're thinking about how this market evolves. We've managed to see the future more clearly than anyone else in the storage market. And we're certainly convinced we continue to hold that edge versus the competition. But we have to be even more careful to not tip our hands given how much scrutiny we're now under. We've got the whole storage industry watching us and trying to figure out what we're going to do next.

Any time I asked EMC CEO Joe Tucci [pictured] that question, he would say, "If I answer that question, then the price of the companies in that space would immediately rise."

That's a good answer.