Pure Storage announces GA for FlashBlade unstructured workload solution

FlashBlade has been in Directed Availability since last July, but it is now broadly available to all customers through the Pure Storage partner channel.

Mountain View CA-based Pure Storage has announced that FlashBlade, their new product for high-performance unstructured workloads, has entered general availability. It is now shipping in both 8.8TB and 52TB blade capacities.

“We first announced FlashBlade a year ago at our Accelerate conference,” said Matt Kixmoeller, VP of Products at Pure Storage. “Since then we executed a beta program with a handful of customers, and in July, we shipped with limited Directed Availability across a wide range of industries, working closely with early adopter customers and getting a good handle on the use cases. It is now available generally.”

FlashBlade becomes Pure’s second product line, following their FlashArray//M offerings, which are for structured data.

“We have been talking about the all-flash data centre for five years, but because customers have both structured and unstructured data, there has been a gap until now,” Kixmoeller said. “Now our partners can offer both structured and unstructured solutions to their customers.”

Kixmoeller said the opportunity in the high end of the unstructured space is vast because there is little in the space at present. The closest thing out there will be Dell EMC’s all-flash Isilon product, which trails FlashBlade slightly on the road to the market. It was announced at Dell EMC World last October, with availability scheduled for some point in 2017, but has not yet hit GA.

Matt Kixmoeller, VP of Products at Pure Storage

“There just hasn’t been a lot of innovation in the unstructured space,” Kixmoeller said. “NetApp and Isilon are the true classic players, and there really hasn’t been a new vendor in this space for a decade, except for some exotic HPC solutions. NetApp is good for small block I/O, and hasn’t been in the high end of the unstructured data market. Isilon is good at large block I/O, although we don’t see any fundamental innovation with their all-flash, just retrofitting with flash drives. What we are seeing now in the market is a need for both to be combined – to mix large streaming I/O with small random I/O together. That’s one of the unique things about FlashBlade.”

Kixmoeller said that FlashBlade’s market falls into two general buckets – technical computing and broad scale analytics.

“Technical computing has been the heartland of NetApp and Isilon and includes use cases like oil and gas and genomics processing,” he stated. “It also involves more technically driven use cases where they understand the value of performance. Analytics is something we believe every Global 5000 needs, especially since they all have data scientists now. It’s a great time for them to look at all-flash.”

The FlashBlade has 1.6 PB of capacity in a small 4U form factor, making it an extremely dense product. Kixmoeller said that its rated tests of 1 million NFS IOPS and 15 gb/s of bandwidth were exceeded in real world use cases.

“One of the pleasant surprises of the early adopter program was system performance,” he said. “We do a lot of testing in the lab, but don’t really know how it will perform with real world workloads until we rest it there. In Directed Availability, the FlashBlade exceeded those numbers.”

Kixmoeller said the Directed Availability work had also given them terrific references.

“One oil and gas customer had a legacy solution with 22,000 hard drives, and we replaced that with one 4U FlashBlade,” he said. “We have also seen good results with companies who build ‘planes, trains and automobiles.’ They do massive simulations of airflows and driving, and FlashBlade can drive much faster simulations.” The Mercedes-AMG Petronas Motorsport Formula 1 team is one early adopter customer in this area.

“We really see the move to Apache Spark from Hadoop as a good linkage to this product,” Kixmoeller said. “The transformation from Hadoop to Spark assists with more interactive real time queries, which makes Spark and FlashBlade like peanut butter and chocolate. The UC Berkeley’s ADAM/Big Data Genomics project would take 12 hours before to do genome sequencing, using a cloud provider for storage. FlashBlade took that to 30 minutes. This much faster analytics capability is seeing new use cases pop up in industries like financial services and web companies as well.”

The strong demand and expanding use cases mean that the growth potential for FlashBlade is exceedingly strong.

“We think that the world is pivoting towards high performance unstructured data, and that this will be a much more dynamic market over the next decade than structured data,” Kixmoeller said. “Of course we believe that only about 15 per cent of the transition to flash has taken place, so the structured data market should still do well.”

Kixmoeller stated that because FlashBlade is simple to deploy, their existing partner base from FlashArray//M will be able to handle FlashBlade deployments.

“There is a very broad set of analytics use cases which will bring us into some new and very specific verticals, and we expect some modest partner additions to go after those verticals, but by and large this will be sold by our same channel as before,” he said. “One of the hallmarks of Pure is effortlessness and we have tried to drive that into the design of both our products lines. They are both managed by the same platform, so it’s a simple deployment exercise. All the existing Pure partners can sell this immediately.”