Violin Memory: This Is The Impact Event Before The Extinction Of Hard Disks

Hailing the arrival of all-silicon enterprise storage and the end of mechanical storage arrays, Violin Memory is rolling out new products that deliver an 80% reduction in cost per input/output operations per second (IOPS) and a 15-to-1 reduction in physical consolidation of hardware from a disk array vendor. The company says its new 6000 series and the addition to the 3000 series of solid-state drives mark the beginning of the end for hard disk drives, equivalent to the impact event that killed off the dinosaurs.

Available in either single-level cell (SLC) or multilevel cell (MLC) flash technology, the 6000 series can be clustered together, scaling to petabytes of storage and tens of millions of IOPS. Targeted at performance needs, the SLC-based systems support 16 Tbytes per array and deliver 1 million RAID IOPS, while the MLC-based systems are optimized for capacity and support 32 Tbytes per array and deliver 500,000 RAID IOPS. The company also unveiled the 3220, which doubles the storage capacity of the 3200 series to 20 Tbytes per 3U shelf. The 6000 series will start shipping in volume in early 2012, while the 3220 is now generally available.

Violin says it can cover virtually all of an enterprise's Tier 0 to Tier 3 storage requirements with all of the 6x0 features required for reliable and cost-effective performance. While it admits disk storage isn't going away any time soon, it believes flash technology can now beat disk across the board. In May, it set a world record dual-socket TPC-E benchmark with HP, and in July IBM scanned 10 billion files 37 times faster than the previous record of 1 billion files in three hours.

"I am seeing an increased use of solid-state storage across the enterprise and expect the amount of solid-state capacity shipped into the enterprise to increase by over 170% this year," says Jeff Janukowicz, research director, solid state drives and HDD components, IDC. "This includes everything from using SSDs in a tiered storage environment, or as a data cache, or even as a complete solid-state solution like Violin's solutions.

He says that for a lot of applications, only a small percentage of the data is hot enough to benefit from solid state, and for these environments, a multitier approach using SSD or a cached SSD environment works well. However, it can be challenging to identify what data is most frequently accessed and would typically require some additional software. "As a result, I am seeing an increased amount of purpose-built, completely solid-state solutions like Violin's. If your data set is a reasonable size--for example, a 5-Tbyte database--putting the entire solution on solid state as primary storage can deliver the benefits of solid state and accelerate application performance without overhead or over-provisioning of resources, which in the end makes them cost-effective."

From a performance standpoint, there is no question Violin can replace Tier 1 storage in the enterprise, states Arun Taneja, founder and consulting analyst, Taneja Group. However, to replace Tier 2 and Tier 3 storage, one has to be competitive on a dollar-per-gigabyte basis (since Tier 2 and 3 are capacity tiers, relatively speaking). "I don't believe Violin replaces those today. But replacing Tier 1 alone is an incredible accomplishment. In fact, there is a set of new problems that can only be solved with these arrays. And, in many cases, it would replace hundreds, if not thousands, of HDDs for incredible reduction in power, cooling, space and cost."

Taneja believes Violin has done a great job in dealing with performance, availability and latency. "Since Violin engineers come from memory side rather than the storage side, they have focused more on these. The other players, such as Pure Storage, Tintri, GridIron and several others, come from a storage heritage, so they have assembled [on top of the array] snapshot, replication and other software that enterprises have come to love. Violin is depending on upper layers [networking, server or application] to perform these functions. This is OK for now since Violin has a lead on performance, availability and latency, but it would behoove them to add these functions over the next year."

In the meantime, the battle lines are being drawn and the next 12 months will see gigantic fights among these players, he says. "This will happen at the same time that big guns like EMC, HP, IBM, Dell and HDS are trying to keep all these guys out of their accounts. Then we will see one big company buy an all-flash array company, and all hell will break loose. I can't wait. I see nothing but goodness in all this for the customer."

See more on this topic by subscribing to Network Computing Pro Reports Parallel Memory Optimized (subscription required).