Pure Storage CEO Giancarlo races toward NVMe-oF, ponders HCI

The new Pure Storage CEO sees NVMe as the immediate future for flash storage; he's also considering hyper-convergence and 'looking at some things' on the M&A front.

When Cisco veteran Charlie Giancarlo took over as Pure Storage CEO in August 2017, the all-flash storage pioneer's short-term goals were to hit $1 billion in annual revenue and finally achieve profitability. Pure said in March it accomplished both of those milestones by achieving $1.02 billion in sales for the last fiscal year and posting a profit in its most recent quarter.

What's next for Pure Storage, which launched in 2009 in the early days of flash storage and went public in 2015? Giancarlo is looking to build on the growth the vendor experienced under its original CEO, Scott Dietzen. Dietzen left that role to hand the reins to Giancarlo, but remains Pure's chairman.

Giancarlo served in various senior executive roles at Cisco from 1993 to 2007 and at one point was viewed as the successor to then-CEO John Chambers. But Giancarlo left instead to head Silver Lake Partners, the equity firm that enabled Dell Technologies to return to private ownership and then helped fund Dell's merger with EMC in 2016.

With Pure's growth on an upward trajectory -- the vendor was cash-flow-positive all last year, with annual revenue growth of 40% -- Giancarlo was brought in to take the company to the next level. "There is no enterprise IT company that got to the $1 billion mark faster than we did at Pure," Giancarlo said.

The Mountain View, Calif., vendor's flagship FlashArray provides block and file storage, including FlashArray//X based on nonvolatile memory express (NVMe) PCIe flash. Pure also sells FlashBlade for unstructured data and plans to introduce a converged AI product this month in conjunction with Nvidia.

Pure Storage CEO Charles GiancarloCharles Giancarlo

In a Q&A, the new Pure Storage CEO discussed his plans for maintaining the company's momentum. Those plans include capitalizing on the coming NVMe wave, exploring hyper-convergence and using some of Pure's cash for acquisitions.

External networked storage has lost some ground to commodity storage, yet Pure grew 48% year over year last quarter. What makes you believe Pure Storage can continue to buck the flat industry growth?

Charlie Giancarlo: If you had asked me a year ago if I would ever be CEO of a storage company, I probably would have said 'no' -- for exactly that reason. When I looked into it, though, I found that while people in storage say the market is commoditizing to white boxes, users see data as the central asset. Furthermore, nobody expects growth in computing to slow down anytime soon. Computing is a three-legged stool made up of advances in processing, advances in communications or networking and advances in storage. If any one of those legs stops growing, you're going to have a bottleneck.

Conventional wisdom is just plain wrong. For computing to advance, the storage blade has to continue to scale and grow. We see ways to make that happen. The bottleneck used to be networking. Now, I believe the bottleneck is storage, or soon will be storage. That gives us a great opportunity.

Scott Dietzen said he stepped down as Pure Storage CEO to bring in a different voice for the next phase of growth. Given Pure's early success, how do you plan to take the company forward?

Giancarlo: I've been fortunate in my career to have experience in a company that underwent hypergrowth and learned how companies need to modify their processes and other things in order to continue to scale. The way you organize or manage in a company of different sizes is just different. I think Scott realized that his enjoyment and his experience were up to the $1 billion range. While I've been in companies of that size, I've also been in a company [Cisco] that went from $1 billion to $45 billion. But Scott is staying on as Pure's chairman, which I think is really important to us.

In the early days of Pure, having an all-flash array alone was a competitive advantage in what was primarily a hard-disk world. That's less of an advantage now that nearly every storage array vendor sells all-flash. What's the key to sustained differentiation as the market gets more homogenized? 

If you had asked me a year ago if I would ever be CEO of a storage company, I probably would have said no.
Charlie GiancarloCEO, Pure Storage

Giancarlo: We designed our software stack from the ground up to make the best use of flash. We get two to three times the data compression of competitors. We're able to give you more capacity in a smaller space, with lower power consumption. We guarantee our flash for life, whereas our competitors are burning out the flash after a year or two.

The headline is that 20% of our sales last year were for NVMe. By the end of this [fiscal] year, the vast majority of our revenue will be NVMe-based. We don't think any competitor has more than 1% of their product line enabled with NVMe. So, we're well ahead. I'm very certain we'll be the first [storage array vendor] with NVMe over Fabrics, but we don't have a date on that yet.

What do you see as the most significant market development for flash storage in this era of convergence and software-defined everything?

Giancarlo: I would say the storage market is really part of an overall converged market taking place. By that, I mean convergence of data. It's a big part of the reason I joined Pure. As communication speeds in the data center reach 100 Gbps or higher, they become commensurate with speeds on backplanes. There is going to be greater convergence of compute, storage and networking, but it's not only the products that are converging. 

Customer expectations are converging, in terms of having a single management pane to simplify management. Convergence is not a cloud unless your data converges. They call it a data center. It's not a compute center, it's not a networking center, and for that matter, it's not a storage center. It's the data that is the valuable entity.

Pure Storage has FlashArray for block storage and FlashBlade for unstructured data, but you lack a hyper-converged infrastructure (HCI) product. Will Pure jump into the HCI market? Is the forthcoming Nvidia-Pure product a step in that direction?

Giancarlo: We have some work to do there. There are varying definitions of hyper-converged infrastructure. Nutanix is doing quite well with its brick-by-brick approach. I've talked to a lot of customers since I became Pure Storage CEO, and their definition is to have a plug-and-play system with simple orchestration to scale compute, networking or storage. Building a system that integrates compute and networking, using RESTful APIs, is definitely an area of focus for us.

With data converging in both primary and secondary use cases, is it enough to just have one tier, as many all-flash arrays have?

Giancarlo: We think convergence also presages new architectures in the data center. The existing view of scale-out was created at time when storage was exploding, but the SAS disk interface between systems was 1 gig. The higher network speeds -- not to mention technologies like NVMe -- give you the ability to access storage as if it were right on the chassis.

People have been thinking in terms of tier one, tier two and tier three for so long that it's become axiomatic with storage. They are used to having to replicate their data multiple times to allow different applications to access it. What we tell customers is, hey, your tier-two data can access primary storage safely, and you can leverage your tier-one investment in flash to replace other tiers. You can tell by the look they give that it's a completely brand-new thought.

Pure has about $600 million of cash on hand. How do you plan to use it? Are you considering acquisitions?

Giancarlo: Cash is there for two reasons, primarily. One is to weather an economic downturn. You always want to know if the economy turns sour that you have the cash to ride through a recession. And it's there for strategic purposes, such as M&A [mergers and acquisitions]. We certainly are looking at some things, but whether one or more [deals] come together in a near-term time frame is so difficult to predict. There is so much serendipity involved.

Dig Deeper on Flash memory and storage

Disaster Recovery
Data Backup
Data Center
and ESG