Home | Storage | X-IO Technologies Takes SPC-1 Value Top Spot with New All-Flash Array
X-IO Technologies Takes SPC-1 Value Top Spot with New All-Flash Array

X-IO Technologies Takes SPC-1 Value Top Spot with New All-Flash Array

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 Filament.io 0 Flares ×

Following on from HDS and NetApp posting news about their Storage Performance Council SPC-1 benchmark results, we’ve seen another two vendors perform tests that have quickly trumped NetApp in the price/performance category.  Current top spot now goes to X-IO Technologies and their new ISE 800 All-Flash Array, which achieves a $0.32/SPC IOP figure, some 40% lower than that achieved by NetApp.  We’ll come back to X-IO in a moment, but as an aside, you have to feel a little bit sorry for Dell, who posted their figures a week later and only achieved second place with $0.37/SPC IOP, although that still beats NetApp by a long way.

X-IO Technologies was spun-out from Seagate as the Seagate Advanced Storage Group in 2002 and went through acquisition by Xiotech before becoming independent again and moving to Colorado Springs (see more history details here).  The company has been selling the ISE or Intelligent Storage Element for some time, based on the ability to more accurately manage the failure modes of hard drives through custom controller microcode.  The product is now in the third generation (G3).  In many instances when a disk fails, only part of the surface of a platter may become inaccessible and there’s no reason why the remainder of the device can’t continue to be used.  The result of this careful drive management gives X-IO the ability to provide warranties of five years on each appliance, with what is known as “zero touch operational simplicity” – you simply install and leave the box to run.

For those of us from traditional storage array backgrounds, this “set and forget” mentality may be a little hard to swallow.  Surely failing drives have to be replaced?  Well, there’s more to getting high availability out of a storage array than simply deploying clever microcode.  Many of the longevity benefits come from hardware engineering to fix some of the common problems experienced with mechanical spinning media.  For more background information I recommend watching the X-IO presentations from Storage Field Day 5, in which some of the specifics are discussed.

XIO-IAFSo what about performance?  How can ISE appliances out-perform traditional arrays and still claim to deliver 100% performance at 100% capacity?  This is where clever software comes in, with features like Intelligent Adaptive Flash (IAF).  IAF allows volumes to either in a hybrid mode (using flash and HDD) through a feature called Continuous Adaptive Data Placement (CADP) or to be pinned to a tier of storage depending on whether the requirements of the volume are either high or low performance (flash or HDD respectively) – a feature called Media Affinity.   The result is that the new G3 ISE 780 can achieve 300,000 IOPS (max) or 140,000 IOPS doing “real world” OLTP workloads.  The all-flash ISE 800 Series achieves up to 400,000/260,000 IOPS and can deliver consistent sub-millisecond latency.

In the past, the achilles heel of ISE has been features.  We all know that going forward, features such as data optimisation and data protection are going to be high on the list of potential purchasers.  With the G3 release, ISE gains thin provisioning, QoS, active/active synchronous mirroring, with a roadmap for additional support in the future.   For virtualisation, there’s support for VMware VASA and for OpenStack a Cinder driver.

 

The Architect’s View

The storage industry is changing.  This statement should come as nothing new to the readers of this blog.  The old guard of big iron with complex provisioning & management is being chipped away by new players with innovative and simpler solutions (don’t take my word for it check out what Barron’s has to say).  In the past, X-IO may have been overlooked because the products were deficient in certain features and to be frank, product marketing wasn’t good enough.  Things seem to have changed for the company and as end users look to adopt hyper-converged models, eliminating storage altogether, then products like X-IO will stand out because they can deliver dedicated storage without the need for storage skills.  The challenge for the future will be in keeping the features coming as the platform evolves.

Related Links

A Langton Blue authored paper on flash is available (for a charge) here – https://www.langtonblue.com/product/technology-report-flash-in-the-enterprise-2015/

Subscribe to our mailing list for more content

Comments are always welcome; please read our Comments Policy first.  If you have any related links of interest, please feel free to add them as a comment for consideration.  

Disclaimer:  I was personally invited to attend Storage Field Day 5, with the event team covering my travel and accommodation costs.  However I was not compensated for my time.  I am not required to blog on any content; blog posts are not edited or reviewed by the presenters or Tech Field Day team before publication.

Copyright (c) 2009-2015 – Chris M Evans, first published on http://blog.architecting.it, do not reproduce without permission.

About Chris M Evans

Chris M Evans has worked in the technology industry since 1987, starting as a systems programmer on the IBM mainframe platform, while retaining an interest in storage. After working abroad, he co-founded an Internet-based music distribution company during the .com era, returning to consultancy in the new millennium. In 2009 Chris co-founded Langton Blue Ltd (www.langtonblue.com), a boutique consultancy firm focused on delivering business benefit through efficient technology deployments. Chris writes a popular blog at http://blog.architecting.it, attends many conferences and invitation-only events and can be found providing regular industry contributions through Twitter (@chrismevans) and other social media outlets.
  • Dimitris Krekoukias

    Hi Chris, Dimitris from NetApp here.

    When looking at SPC-1 results for SSD systems it’s important to look at several things:

    – Whether the configuration was heavily discounted or not

    – The latencies achieved

    – The performance at a low latency point. Since that’s much of the point in acquiring an SSD system…

    The new XIO and Dell results are both discounted by 45%, which improves their posted $/IOPS number. Our results are not discounted, a 45% reduction on the price of the EF560 would propel it to the top of the chart again.

    Latency-wise, the Dell scored over 4ms it looks like, and the low latency results are very low.

    XIO scored better but at 500 microseconds its performance is also not as good as the EF560’s.

    Some analysis here: http://bit.ly/18oWI1R

    Thx

    D

    • http://architecting.it Chris M Evans

      Dimitris

      We only have your opinion on the pricing. The EF560 SPC figures don’t show list/discounted values, plus I can’t find an EF560-0001-R6 on any pricing lists I have checked.

      • Dimitris Krekoukias

        Chris,

        It’s not my opinion :)

        There was no discount applied to the EF560 SPC-1 offering. If you look at the SPC-1 Executive Summary for the EF560, page 8, you can see the offering was submitted at list pricing – no discount applied. If a discount is applied you would see an additional column, it’s not there for the EF560 entry. Indeed, vendors need to disclose discounts.

        As for the part number, that hit our price lists this past Monday and should be there now.

        Having said that, even at list price (I did the math), the XIO does remain a bit less expensive than our system, until you look at its performance at very low latencies, where it doesn’t perform as fast, which skews $/IOPS again in our favor.

        Attaching a bitmap of my spreadheet where I analyze this (will update my blog post shortly).

        Thx

        D

    • Gavin McLaughlin

      Dimitris, Thanks for posting the NetApp analysis but this is before both the Dell and X-IO results were published so I’m not sure why this is here for any other reason than blatant advertising? Are you going to update that page to include the other results.

      SPC are very clear that vendors should show “Street pricing” – why didn’t NetApp do that then? Do you always tend to sell at list price? if the submitted price really was at list then that seems a little naive.

      I’m sure we can all pick holes in each others’ benchmarks as you did however the point of these tables is to show leadership in price/performance and right now, X-IO are in the number one spot which we’re very proud of.

      I’m sure customers are intelligent enough to use these benchmarks as guidelines rather than a sole research point and I’d absolutely love someone to do a head to head comparison of the NetApp and X-IO boxes.

      Incidentally, what’s the warranty included on that NetApp config? Only the X-IO one is five years as standard – I’m sure that will make a pretty big difference if everyone did the same on SPC-1 benchmarks.

      • Dimitris Krekoukias

        Hi Gavin,

        Check the article – it’s been updated with all the latest results (maybe you read it before I updated it).

        Interestingly, NetApp remains the leader at low latency price/performance (the whole point of my SPC analysis articles is to show people how to interpret the results – for instance, Dell’s result is at over 4ms latency, and yours at over 2ms – anyone not looking at the detail will miss that and just focus on the published summary).

        And no, we don’t sell systems at list price, nobody does, but we’ve always submitted list price for SPC-1, historical thing I guess (plus we sell to astronomically more customers and more types of customers than you guys so people get different discounts, based on volume for example).

        Thx

        D

        • Gavin McLaughlin

          Even at sub 1ms latencies, the X-IO price-performance numbers are better. If you’re focusing in on 0.5ms latency and IOPS then NetApp wins – woo hoo – how many people need 0.5ms latency exactly?

          As I said, customers have choice and the whole idea of SPC-1 is to give third party corroboration of proof of claims. I think we’ve both achieved that and maybe we should focus on those that can’t give proof of claim such as Pure and the de-dupe factoring game vendors?

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 Filament.io 0 Flares ×