NetApp learns from past integration failures to leverage SolidFire

Twelve years ago, NetApp acquired scale-out NAS startup Spinnaker, and began a long and horrific series of integration problems. This year, they acquired SolidFire’s scale-out technology, and took steps to learn from the lessons of the past.

val-bercovici-300

Val Bercovici, SolidFire’s CTO at NetApp

NetApp sees its acquisition of SolidFire, which closed last February, as a key factor in its transition to a next-generation datacentre. Successful integration is the key here – something that NetApp has not always accomplished in the past. To that end, they have mapped out a process to try and ensure that everything goes much smoother this time around.

“The biggest legacy of cloud so far is that it has shown a different model for legacy infrastructure,” said Dave Wright, SolidFire’s Founder, who is now Vice President and General Manager of the SolidFire business at NetApp. “I started SolidFire because at Rackspace, I saw how Amazon changed how people expect things to act. People stopped asking ‘where’s the ticket’ and asked ‘where’s the API.’ You can’t manage and deploy the same systems you’ve deployed for 20 years and call it a cloud.”

SolidFire makes all-flash, scale-out storage solutions, which it sold primarily to the service provider and large enterprise markets. It was designed to be extremely simply to manage, which makes it different from – but complementary to – NetApp’s storage.

“In a world where demand is growing quickly, or changing unpredictably, or both, the traditional model of building out an infrastructure to run an app doesn’t work any more,” Wright said.

On the other hand, Wright acknowledged the change, in the evolution to a new generation data centre, will be a long one.

“The move to the next-generation data centre is much bigger than the shift to flash, and involves much more than storage,” he said. “It will be a decade-long transition, at least. Because customers won’t that make leap to the next-generation data centre overnight, they will need a bridge from the old to the new — and SolidFire is great for that.”

Executing on this vision requires a smooth integration, however, and NetApp’s record in the past here has not been the best. In particular, the acquisition of Spinnaker Networks, which closed in early 2004, comes to mind. Spinnaker also had an innovative next-generation architecture, but the integration was extremely problematic.

“We learned from the Spinnaker acquisition,” said Val Bercovici, a longtime NetApper who moved over to the SolidFire side as CTO following the acquisition. “The engineering integration there was bad. We had naïve intentions of integrating the Spinnaker scale-out technology into ONTAP – and you could do that – but it took 8-10 years longer than we thought it would.”

Making sure things go very different this time is a key part of Bercovici’s job.

“I volunteered for the role,” he said. “I spent eight years in the North American office of the CTO directing investments, where I identified trends. When we announced the intention to acquire SolidFire, I saw the opportunity to apply in practice the theory I had spent so much time on before I knew the SolidFire team and seen the product. In 2010, when I first saw the concept, I thought it was too ambitious. But over the years, I saw the progress and developed a respect for the product before it became part of our portfolio.”

SolidFire wasn’t acquired to ramp up NetApp’s all-flash portfolio. Their own all-flash solutions were already moving up in the market. But the deal was widely seen as a flash play – in part because Bercovici said NetApp gave that impression, by focusing on the underlying technology rather than its core objective.

“We got our positioning wrong in February,” he said. “We said it was a third flash product in our portfolio. “It’s enabled by flash, but it’s not a flash product. It enables an AWS elastic storage experience on-prem.”

Co-ordinating that capability with NetApp’s other product lines will be critical going forward. Bercovici stressed that ONTAP and SolidFire won’t be integrated. The plan is to have the two operating systems co-exist long term, both to leverage the SolidFire brand equity and the two systems’ different use cases.

“The engineering teams have been fully integrated and there’s a lot of coordination between the technical sales teams,” he said. “We are creating an office of the CTO that’s very modern in that context. We will be a liaison with other engineering teams, with ONTAP, E-series, StorageGRID, AltaVault and OCI.”

That should mean effective cross-pollination between the different groups.

“One of my earlier research projects was looking at our core replication, SnapMirror, and modularizing it, so that secondary targets didn’t have to be ONTAP,” Bercovici said. Adding that logical replication layer of abstraction allowed us to integrate AltaVault, E-series, and now, SolidFire.” More file services will also be coming on top of SolidFire over time.

The channel program integration has been completed, a process that wasn’t massively complex because SolidFire’s old channel was fairly small, with under 50 partners.

“Their goal was to get close to 100 per cent channel fulfilment through the channel,” Bercovici said. “Their product was designed by cloud datacentre people, and their early customers were datacentres, so even through the product itself was simple, it was usually a high touch sale because the competitors that had to be replaced were fairly entrenched.”

NetApp is training their own legacy partners on SolidFire, and has plans to broaden out their Total Addressable Market.

“Traditionally, SolidFire sold to service providers – a small majority of sales – and digitizing enterprises,” Bercovici said. “With NetApp, we are recognizing that the digital enterprise market is bigger and have focused on that enterprise opportunity. We are also working on scaling it down to the midmarket.”

Bercovici acknowledged the mid-market is probably a year or so away, however.

“With a minimum bundle of four nodes to start, you are looking at an introductory price of $USD 75,000,” which is a little steep for the mid-market. We don’t think with our current architecture we can get below that, but we hope to change that by this time next year.”