Pure Storage Release 4th Generation of FlashArray Hardware

Pure Storage Release 4th Generation of FlashArray Hardware

Chris EvansPure Storage, Storage

Last week Pure Storage released the 4th generation of their all-flash FlashArray branded hardware.  This saw the introduction of two new models and a range of new software features in the parallel release of Purity 4.0, FlashArray’s operating system.

Hardware

The FlashArray platform now has three generation 4 models, the FA-405, FA-420 and FA-450.  The previous FA-400 model is renamed to FA-420, with no change to the specifications.  FA-405 is a new entry-level array based on dual 1U controllers.  FA-450 delivers higher-end capacity of 70TB (raw) and up to 250TB usable.  The “usable” figure is calculated on an assumed 6:1 ratio of data reduction technologies after data protection (RAID) overheads are taken into consideration.  Full details can be found in the FlashArray specification sheet.

Software

Purity (the software running FlashArray) is upgraded to release 4.0 and introduces FlashRecover, which brings data replication features to the FlashArray platform.  Replication is implemented through snapshot copies, where snapshots are created locally and shipped to a remote array.  The first release of FlashRecover will provide snapshot RPO at 1 minute intervals, with the promise of lower RPOs in the future.

There are a number of ways to implement replication and in this instance Pure have implemented the asynchronous model with time-based data transfers.  This scenario is probably the least impacting on performance – synchronous replication affects host I/O performance due to latency and replicating every block (in a near CDP model) can have throughput issues without adequate bandwidth.  Using snapshots to replicate data means high-intensity write data will be consolidated into a single snapshot reducing the bandwidth required.  Of course the tradeoff here is that block-level granular recovery isn’t available, but replication on top of de-duplication can be highly efficient as existing data blocks on the target array only generate a metadata update.  This means we may see a highly granular level of replication in future releases.  More details on FlashRecover can be found here.

In addition to FlashRecover, Pure have added REST API support to Purity 4.0, delivered Windows 2012 VSS provider support and implemented FlashProtect, a data locking feature specifically for customers in the “intelligence community”.  Read into that what you will!

The Architect’s View®

Generation 4 of FlashArray extends the reach of the platform to new higher and lower end customers and use cases.  It’s worth remembering that in many cases, Pure is not targeting other flash array vendors products but looking to take out existing traditional hard-disk based arrays and so having a wider supported capacity range – with upgrades in place – will improve this ability.  However, in the long term, the FlashArray architecture is still a dual controller configuration based on a scale-up design.  We are seeing a move to scale out in compute and storage, so scale-up will have a limited lifetime.  To survive in what is becoming an increasingly competitive market, Pure will need to evolve to other storage designs and architectures.  With so much money in the bank (and a lack of the technical founders at launch events), I can only conclude that this work is under way and we’ll see the results of this labour in due time.

Images copyright (c) Pure Storage.

Related Links

Copyright (c) 2007-2023 – Post #3C8E – Brookend Ltd, first published on https://www.architecting.it/blog, do not reproduce without permission. Photo credit iStock.