Industry vet Lewis reveals plan to save Violin flash storage

Mark Lewis said when he became CEO of Violin Systems a few months back, a friend asked, "Why are you joining Violin?"

The friend knew Violin's history. Lewis was replacing ailing Ebrahim Abbasi as CEO, taking over a company that had been through several chapters already -- including Chapter 11 bankruptcy. Violin emerged from bankruptcy in 2017 when private equity firm Quantum Partners put up $25 million to satisfy Violin's outstanding debt and fund restructuring.

Lewis first joined Violin as a consultant to Abbasi, following the demise of Lewis' venture capital-backed Formation Data Systems. Formation Data couldn't get its software-defined storage product to market fast enough to satisfy investors. When Abbasi stepped down for medical reasons, Lewis took over as Violin's CEO.

"We are effectively a startup," Lewis said of Violin.

Violin went from a startup in 2009 to a public company in 2013, fueled by early success as a flash storage pioneer. But the rest of the storage field jumped into all-flash and both startups and established vendors surpassed Violin sales.

Lewis' experience as a storage executive goes back to 1990, when he became general manager of Digital Equipment Corp.'s storage OEM business. He followed with executive positions at Compaq, Hewlett-Packard and EMC before founding Formation in 2013.

Lewis has recruited other industry veterans to Violin. Rick Ruskin is the senior vice president of worldwide field operations, after occupying a similar post at Kaminario. Gary Lyng, vice president of product management and marketing, previously worked for SanDisk, and his resume includes stops at EMC, NetApp, Hewlett Packard Enterprise and Veritas.

We spoke with Lewis about why he joined Violin and his strategy for plotting its comeback.

Given Violin's ups and downs, how challenging has it been to maintain the confidence of existing customers and win new Violin flash storage accounts?

Mark Lewis

Mark Lewis: I had a sales friend tell me 'Congrats on [joining] Violin, but why are you joining Violin?' Everybody has seen the press about the bankruptcy and thinks things here must be terrible. Clearly some bad execution happened to drive us there, but that all helped to battle-harden the technology. I'm excited by the potential we have.

One of the things that impressed me most when I got here was the incredible IP [intellectual property] Violin has. The IP was in great shape, but we didn't have a lot of process. Cuts had been made to the sales team. We had no growth engine because we had no marketing and sales. Now that we have private equity funding, we have rebuilt a channel partner-oriented sales team. We're rebuilding relationships with existing customers, and we already have 20 new logo opportunities [for new customers] this quarter.

We still have a lot of large installs. About 150 companies of the Fortune 500 currently use Violin to run their most mission-critical applications. Now that we're back to being privately owned, we can leverage our IP assets and installed base to re-establish ourselves on a deeper footing. We're not doing cloud tiering. We're not doing object stores. We're not even doing file storage. We're sticking with block storage, Fibre Channel, iSCSI Ethernet and our approach of selling only through the channel. We think Violin flash becomes a complement that can coexist within your environment.

Why didn't Formation Data Systems survive, and how will those lessons help you in leading Violin and succeeding with Violin Flash Storage?

Lewis: They are two products in completely different markets. At Formation, I advocated for a certain portion of the market to move to software-defined storage for midperformance or bulk storage. It's a valid market, but what we learned at Formation is that there isn't a lot that's broken in the classic storage market. People in storage tend to be conservative. I believe software-defined storage ultimately will be successful, but it won't happen quickly. It will be a multidecade transition.

The core weakness of the original Violin flash storage technology was that it was built as very fast storage hardware, with no software. That made it a struggle for enterprises to adopt it. It was a hard lesson to learn. Obviously the company struggled for a long time getting that right.

The good news is we've now got a platform. We have battle-tested software capabilities that we're building on, and we will evolve to using more off-the-shelf commodity hardware, like NVMe [nonvolatile memory express], and continue to move to more of a software setup. But being in the enterprise, we aren't trying to be an all software-defined storage. To address the super-high-performance environments, you have to integrate hardware and software to work tightly together. Loose coupling works fine, but we want to beat everybody on performance. We're always going to have a close tie to hardware.

What's your strategy to help Violin avoid the pitfalls that sidetracked it before?

One of the things that will make us successful is improving our focus on ultra-high-performance workloads. People categorize the market now [as] ... all-flash arrays, but that's really not a market.

Lewis: One of the things that will make us successful is improving our focus on ultra-high-performance workloads. People categorize the market now in the general term of all-flash arrays, but that's really not a market. It's something that can be counted, but all-flash arrays don't make up a buying choice in any particular way. We believe there's a lot more depth to it than that.

The total addressable market for all-flash arrays last year was about $8 billion. Violin isn't even a drop in that bucket. We're effectively a startup. Violin is known for incredible IP around performance. We're focusing just on the segment of the market that needs extreme storage performance. We're not going after the generic market to compete with Pure Storage, NetApp, Dell EMC or others. We'd be very happy to carve up a billion dollars of the high-performance portion of the market.

Are investors asking about when you'll be in the black? Do you have a projected timeline for turning a profit with the Violin Flash Storage platform?

Lewis: We're taking a conservative, cash-oriented approach. We aren't going to focus on profitability now, but on generating positive cash flow. That has to come first as a way to sustain the business. I believe we're about two years away from being able to do that. But we won't have to be a billion-dollar company to have positive cash flow. Thankfully, we have IP that keeps our costs and margins in good shape, and we're able to exploit flash for performance pretty cost effectively.

Other vendors are trying to play a game of replace an old disk array with flash. It's a big market, but the negative is a race to the bottom on cost. We're attacking the market from the top down and have a lot of runway to [achieve] profitable growth, without worrying about cannibalization.

What is Violin's roadmap for supporting NVMe flash?

Lewis: We will release NVMe over Fabrics [products] this year. The front-end NVMe [support] is important to our customers and will give us improved performance, even above what we can deliver now.

On the device side, it's a different story. We never used SAS or SATA drives. We created our own controller chip that's actually faster than NVMe today. Because of our hardware-software integration, we're working with the NVMe device suppliers to add the software hooks that we need to make NVMe fast enough to go in our Violin systems. We want to embrace that commodity hardware, but we need software improvements to do it. NVMe, as it is now, is actually too slow for us. We have to help NVMe device makers to get caught up with us.

For storage class memory, we're investigating 3D XPoint and other RAM technologies. Our strategy is to give our customers the option to use those technologies as soon as they become commercially viable.