Pure Storage introduces petabyte-scale flash storage for cloud
Matt Kixmoeller, VP of Product, Pure Storage

Pure Storage introduces petabyte-scale flash storage for cloud

Pure Storage has recently introduced the availability of petabyte-scale storage for mission-critical cloud IT, anchored by the release of the next-generation of FlashArray//m the company’s flagship all-flash storage array, which now delivers performance with the simplicity and agility of public cloud. The FlashArray//m now offers customers petabyte-scale capacity with a measured 99.9999% availability and always-on Quality of Service – virtually eliminating storage downtime and application performance risk.

With this release of the next-generation FlashArray//m, Pure Storage is equipping customers with Smart Storage, storage that is purpose-built to be effortless, efficient and evergreen – and capable of delivering the simplicity, automation, resiliency, and customer friendly business model which are all essential for cloud IT.

“Storage underlies everything we do in today’s digital economy, an economy that is increasingly powered by cloud IT,” said Matt Kixmoeller, VP of Product, Pure Storage. “Pure Storage has built the smart storage platform to deliver cloud IT – whether it be public cloud, SaaS, private cloud, or a mix of all three.  Efficient all-flash storage, now at petabyte-scale, is the agile foundation of tomorrow’s clouds.”

Petabyte-Scale FlashArray//m

The newly updated FlashArray//m, the fifth generation of Pure Storage’s flagship FlashArray product, now scales up to 512 terabytes of raw flash, which translates to approximately 1.5 petabytes of effective capacity – all in just 7U of rack space. This scale and density enables customers to consolidate racks of legacy disk storage down to 7U or less.

Leveraging Pure’s EvergreenTMarchitecture, the existing FlashArray customers of every generation can seamlessly upgrade to the new FlashArray//m, allowing customers to take advantage of rapid technology advances in both compute and flash. The new FlashArray//m offers four different controller options to meet a variety of performance and capacity needs:

  • //m10 (introduced earlier in 2016) – up to 25 TBs effective usable capacity (5 – 10 TBs raw storage) and up to 100,000 32K input/output operations per second (IOPS),
  • New //m20 – up to 250 TBs effective usable capacity (5 – 80 TBs raw storage) and up to 200,000 32K IOPS,
  • New //m50 – up to 500 TBs effective usable capacity (20 – 176 TBs raw storage) and up to 270,000 32K IOPS,
  • New //m70 – up to 1.5 Petabytes effective usable capacity (42 – 512 TBs raw storage) and up to 370,000 32K IOPS.

Across these new controller options, the FlashArray//m delivers a 20 to 30 percent performance boost and a 100 – 276% capacity boost over the previous FlashArray//m generation.  Object scale has also been improved by 10X, enhancing overall system scalability and enabling large-scale Copy Data Management with up to 50,000 space-efficient snapshots on the system.

In its first year of shipment, the FlashArray//m has achieved 99.9999% availability across the installed base, which equates to only 31.5 seconds of downtime on average per year, demonstrating the FlashArray//m can deliver the level of resiliency needed for mission-critical enterprise applications and always-on clouds.

Pure Storage is also redefining the way availability is measured and reported to align directly with customer expectations and experience of availability. FlashArray//m’s measured 99.9999 availability is inclusive of upgrades and maintenance, across both hardware and software, and without requiring a second array configured with replication.

Key to delivering this level of availability is not only the resiliency architected into the FlashArray//m, but also Pure1 Global Insight, Pure’s cloud-based predictive analytics platform. Since its introduction almost a year ago, Pure1 Global Insight has helped to automatically identify and resolve thousands of issues, and prevented over 170 Severity 1 issues to date. Pure1 Global Insight is a key ingredient for delivering an always-on experience to Pure Storage customers, as the install base continues to grow.

Always-On QoS

As customers consolidate dozens or even hundreds of applications, concerns about “noisy neighbours” – applications that starve the performance of other applications on an array – can emerge. Typical QoS implementations to prevent noisy neighbours are complex, requiring per-volume min, max, and burst IOPS to be set and then constantly adjusted as application needs evolve. Improper QoS configuration can cause more harm than good!

The FlashArray//m’s new Always-On QoS feature requires zero configuration, so all customers can benefit without the complexity of legacy QoS implementations. Any application can burst up to the maximum utilisation of the FlashArray//m, as long as it is not impacting other applications. If the application becomes a “noisy neighbour” and begins to impact other applications, Always-On QoS will throttle the problem application down to ensure all other applications get their fair share of array performance. Architected with a sophisticated IO credit system under the covers, Always-On QoS is simple and autonomous, making it ideal for most use cases, and builds the foundation for future policy-driven QoS extensions.

NPIV support for host-transparent upgrades

Network Port ID Virtualisation (NPIV) support on the FlashArray//m makes the non-disruptive software upgrade experience even more robust and invisible to the host and server and virtualisation admin teams. With NPIV, IO intended for ports on a FlashArray//m controller that have been temporarily restarted during the software upgrade process is transparently directed to the other FlashArray//m controller, without any dependency on host multipathing software. NPIV removes the risk of downtime due to misconfigured host multipathing software – a common worry for large enterprise environments – and makes storage software upgrades transparent to server administrators by eliminating IO error alerts that would otherwise be generated at the application host layer.

Click below to share this article

Browse our latest issue

Intelligent CIO Middle East

View Magazine Archive