Violin Systems prepares NVMe launch, but not for drives
Violin Systems will make its first platform launch since its resurrection when it brings out a flash array with NVMe support on the front end for host connectivity in September.
Violin CEO Mark Lewis said the vendor doesn’t need end-to-end NVMe yet because its custom built flash modules are faster than NVMe drives now.
The vendor’s flagship array is the Flash Storage Platform 7650, launched three months before Violin went into bankruptcy in December 2018. Private equity firm Quantum Partners acquired Violin’s assets in April 2017 and brought it back to life as Violin Systems. There were tweaks to the platform since then, but no new products.
Lewis said another platform is about to drop. NVMe may not be the main focus, but it will be included.
“NVMe doesn’t help us,” Lewis said during an interview last week at the Flash Memory Summit. “We’re faster than NVMe now, we don’t have a problem. The vendors using SAS and SATA drives, they have a problem that NVMe helps.”
Lewis expects to add NVMe to Violin’s custom chips in early 2019 and switch to off-the-shelf NVMe SSDs when the speed improves in late 2019. But for now, the new Violin platform will keep its custom flash modules.
“We will support NVMe on the front end in the platform we launch next month,” Lewis said. “On the back end, we already have our own storage. ATA and SATA were way too slow, so we built our own controller, FGPA, all of that. So right now, we are quite a bit faster than NVMe on the back end because we soldered the NAND into the board. We’re working with NVMe SSD suppliers to speed them up.”
Lewis said his plan is to sell Violin’s pure performance to customers who need it most, and leave the general flash storage market to the Dell EMCs, Pure Storages, IBMs, NetApps and Hitachi Vantaras of the world. He said the desire to compete across the board doomed Violin the first time around.
“Violin lost that focus,” he said of the flash array pioneer’s previous struggles. “We’re just going to focus on the Tier 0 performance low-latency space. That’s what we do really well, that’s what we do better than everybody else by a long shot, and there’s a market for that.”
Lewis said many Violin’s customers stuck with its storage during the bankruptcy period because they couldn’t find an alternative that gave them enough latency for use cases such as transaction processing. He said Violin has about 100 customers, including five new customers last quarter.
“We were very happy with that,” he said. “We’re like an A Round startup now, with a great customer base. The hardest thing for a startup is getting that early customer adoption, and we already have a lot of Fortune 500 accounts. And we’re back up to about 100 employees, counting contractors.”
Violin Systems didn’t make Gartner’s Magic Quadrant for all-flash arrays released last month. The latest Magic Quadrant included 12 vendors, with seven among the leaders. Lewis said he was happy to be excluded.
“Product sales went to zero last year, so you can argue we were too small [for the Magic Quadrant],” he said. “But I also told Gartner I do not have a strong desire to be on that quadrant. We’d become a niche player. The Gartner Magic Quadrant goes to execution about this massive all-flash array market, and it becomes a scale game.”
Lewis pointed to Tintri, which appeared in the Magic Quadrant as a visionary despite following Violin into bankruptcy, as proof that the Magic Quadrant inclusion does not guarantee success.
“Tintri’s core issue was similar to Violin’s,” he said. “They jumped in the deep end. Every one of their deals had Pure and Dell and HPE, and everybody else in it.”