FlashBlade from Pure Storage, a converged platform for big data workloads, was first introduced in March of 2016 and began shipping to select customers the following July. This week, the Mountain View, Calif.-based storage company announced that it is now generally available worldwide.
Evoking the compute prowess of a blade server, FlashBlade uses the object storage system in the company's Elasticity Scale-Out software, enlisting the system's NV-RAM (non-volatile random-access memory) as a data processing compute engine.
In terms of storage performance, FlashBlade can hit 500,000 input/output operations per second (IOPS). A software-defined 40Gb Ethernet fabric provides a total network bandwidth of 320Gbps within the chassis.
Blades are available in 8.8TB and 52TB capacities. At the latter capacity, a fully-occupied 15-blade system can deliver up to 1,607TB of usable capacity in a 4U system.
"FlashBlade unlocks new capabilities and use cases that are simply not possible with today’s existing solutions by delivering real-time analytics at massive scale and blazing speeds to enable the future of innovation," said John Hayes, cofounder and chief architect at Pure Storage, in the company's initial announcement last March.
In the months since the company launched the FlashBlade Early Access program, making the system made available to a limited number of customers, Pure Storage discovered those claims were true.
"In particular, we have learned that FlashBlade excels in analytics — both classic application suites like Oracle RAC running data warehouses, and the emerging Big Data 2.0 stacks running large scale Apache Spark clusters for iterative queries, machine learning or SQL query processing," wrote Par Botes, vice president of FlashBlade at Pure Storage, in a blog post.
So far, customers include the Mercedes-AMG Petronas Motorsport Formula 1 team, CUProdigy, a provider of cloud-bases software for credit unions, and the Genetics and Genomics department at UC Berkeley. In the latter's case, Pure's hardware is helping to massively slash the time it takes to conduct potentially life-saving research, noted Botes.
"They are using FlashBlade to conduct complex analysis and run data-intensive visualization in 3D. The results have been astounding — queries on Apache Spark that took 12 hours now take just 30 minutes," he said. "Ultimately, UC Berkeley is conducting this research to adjust clinical diagnostics in real-time, which in turn are improving patient outcomes and saving lives."
FlashBlade is available now through Pure Storage's channel partners.
Pedro Hernandez is a contributing editor at Datamation. Follow him on Twitter @ecoINSITE.