While many customers are still on the fence about NVMe adoption in its early days, Pure is making it easier for them to buy FlashArray//X now and upgrade later, by making the upgrade path simple, and changing pricing policies to encourage people to buy //X now.
SAN FRANCISCO – A year ago, Pure Storage introduced their FlashArray//X, their venture into the NVMe space designed to mainstream a technology which everyone acknowledged had enormous potential, but which had still not really taken off commercially. Now, at their Pure Accelerate event here, they are announcing a commitment to the all-NVMe FlashArray//X as the focus of their business going forward – even as they acknowledge that NVMe is still in its early days,
“Today, we are going much, much deeper,” Matt Kixmoeller, Pure Storage’s VP of Strategy, announced in the opening keynote. “We are all-in on Shared Accelerated Storage, with a new family of Shared Accelerated Storage that can bring this to every workload in your data centre. This year, we are all-in on FlashArray//X – not the SATA-based FlashArray//M.”
FlashArray//X leverages NVMe and NVMe-over-Fabric, executing on Pure’s announced vision of a fundamentally more efficient data-centric architecture. The result is an array that makes everything faster – databases, virtualized and containerized environments, test/dev initiatives and web-scale applications.
“We’ve been innovating with NVMe for years,” said Bill Cerreta, VP Platforms at Pure Storage. “At that time  when we introduced FlashCare, we couldn’t talk directly to the flash, but it is how we could use standard SSDs when our competitors had to use expensive enterprise SSDs. We shipped the first dual ported NVMe device in the industry with FlashBlade, managing all the flash globally, giving us incredible efficiency and control. That was something that was unique to Pure. Now we are completing the picture with NVMe-over-Fabric.”
Pure’s approach to NVMe is also fundamentally different from industry competitors, said Chadd Kenney, Pure’s CTO for the Americas.
“Every other vendor leverages intelligence in the drive,” he said. “We wanted to democratize NVMe with intelligence that sits above it. So we built a dumbed-down drive with no intelligence with no intelligence, which is managed by that global intelligence. We believe that this is more efficient, because you won’t have conflicts with the intelligence in the drives.”
The FlashArray//X family features five configurations, from the NVMe-ready //X10 for small application deployments at the low end to the top-of-the-line FlashArray//X90, Pure’s densest, fastest FlashArray.
“As we look forward to Storage Class memory and QLC [Quad-Level Cell NAND flash memory], we are in a great position,” Cerreta said. “The //X90 is the fastest, densest array we have ever built. The controller can also run both NVMe and flash, which will make upgrades super easy. All the new models, from the //X10 to the //X90, are NVMe-over-Fabric ready.”
The //X90 has up to 3PBs of effective capacity in a 6U form factor, delivers latency as low as 250 µs and offers up to a 2X performance improvement over Pure’s FlashArray//M all-flash array.
“Even if you are not yet ready to upgrade to NVMe, this box screams because of the low latency,” Cerreta stated.
The //X70 and //X50 scale to 1.3 PBs and 650 TBs of effective storage respectively. The //X20 and //X10 – Entry-level NVMe-ready arrays for smaller capacity and remote office needs. These come configured with SATA flash by default, but feature controllers fully-capable of supporting NVMe DirectFlash modules out-of-the-box.
In addition, an updated version of the born-for-flash Purity Operating Environment, version 5.1, is available as a non-disruptive upgrade for all FlashArray//M and FlashArray//X customers and features up to 20 percent better data reduction, further reducing the amount of storage customers need to buy and manage. It adds new Pure1 VM Analytics that enable full-stack visibility into VMware vSphere environments, and expands the Pure1 Global Dashboard to enable fleet-wide visibility of all FlashArray and FlashBlade arrays globally on both the Pure1 SaaS portal and the Pure1 Mobile app. Pure1 Workload Planner has also been expanded.
Pure is well aware that the NVMe market is still developing and that many customers will want to wait a little longer, which is why they are emphasizing the easy upgrade path built into the FlashArray//X.
“NVMe is still pretty immature,” Kenney said. “The only really resilient driver for it today is in Linux, for example. The immaturity of the space makes us concerned about diving in too deep, which is why we are testing the market and making it possible for customers to upgrade later. We think though that our ability to combine both SAN and DAS storage in Shared Accelerated Storage will help us expand our TAM significantly as this goes forward, because it allows us to attack the DAS-based solution set. Traditional SAN-based environments didn’t have the ability to do that, because you had to change the software stack.”
Pure has given strong emphasis at Accelerate to their new 0 per cent pricing strategy, which is a clear encouragement for customers to future-proof themselves with FlashArray//X rather than the older FlashArray//M, because the //X will be sold at the same price as the //M.
“With no premium for //X over //M there is no reason now not to go all NVMe,” Kixmoeller said.
FlashArray//X is orderable today for both new arrays and upgrades and will ship in early June. Purity 5.1 is available today. Pure1 VM Analytics, and the Workload Planner and Global Dashboard updates will be made available soon.