Is IBM XIV still relevant?

View Comments

• I'm hearing that >70% of XIV systems sold, have been sold into EMC strongholds. I'd say that statistic alone makes them pretty relevant.

  Competition with IBM's other products definitely happens, but it is clearly not happening the majority of the time. The XIV sales team (still a separate, unintegrated sales organization within IBM) is made up primarily of ex-EMCers...with very good relationships in EMC accounts. Hence the strong penetration into those accounts.

  I can also concur with your observation that the XIV guys tend to take the view that if you're not 110% on board with the product, forsaking all others, that you're an XIV hater. It's a juvenile attitude, but somewhat understandable, given where they came from!! ;)

  One more thing that deserves a comment... Every time someone outside of IBM mentions the success of the XIV, it's a safe bet that trial systems and non-revenue "sales" are going to be mentioned in the same breath...as you did in your post. It was fair a year ago...but IBM has all but completely stopped doing anything like this for the past year. If an IBM business partner or distributor isn't willing to take all the risk, it ain't gonna happen. And it doesn't need to happen any more...the XIV isn't new any more. IBM is having a hard enough time keeping systems in stock to sell to paying customers...they definitely aren't giving them away.

• XIV matters because it's a bet that future large-scale storage systems will be architected using proprietary software running on commodity hardware. That's a tough message to swallow if you're a traditional storage developer building proprietary hardware. There are man-years of development, tens of millions of R&D dollars and hundreds of millions of gross margin dollars at risk if systems like XIV obsolete the need for proprietary storage hardware.

  From a market view, commodity hardware solutions like XIV matter because they address the needs of the fastest growing segments of the storage market, namely rich media, hosting infrastructure, medical image repositories and disk-based backup solutions where scalability, ease of use and lower cost matter most.

  Look at history. A very similar battle unfolded when x86-based servers took over the volume segment of the server market in the late 90's. Sure, RISC systems were full of elegant proprietary hardware and had a dominant installed base, but x86 systems were "good enough" once a stable OS became available with NT 3.1 (ok I'm dating myself).

  This is an architecture war right up there with the open source versus Microsoft enthusiasts. It will be tough for users or vendors to sit on the fence. I'm not surprised that it's a bit contentious even inside of IBM. Expect that the rhetoric from conventional suppliers against these XIV-like systems will increase until it sounds like what we used to hear from SUN Microsystems - back when they mattered.

• Interesting discussion here as well as over in twitterville.

  Lee based on your point that it’s about architecture which I see some aspects to, then that says the IBM DS8000 is part of that same approach of using clustered off the shelf components with custom proprietary software to glue it all together?

  Now that might be a stretch per your points, however then why don’t we see vendors that tout leveraging open off the shelf hardware platforms actually do just that, other than for the drive enclosures or back-ends? Im not saying that the DS8000 is a cluster or grid, simply using it for discssion points as there is enough cluster and grid confusion.

  Concur that x86 and standard processors are the norm and rallying point along with accompanying off-load engines as needed as evidence in the recent Sun/Oracle Exadata 2 which still included a PCIe SAS RAID adapter card. This is probably more of an example of how a custom board does not need to be laid out, rather complimented by off the shelf adapter or off-load cards when and where needed.

  Funny thing however was a conversation where the perception that ASICs were dead, yet, a quick look around the room revealed asics everywhere, just very transparent complimenting the off the shelf technologies. Thus for some applications, good is good enough, for others, there needs to be more than just good enough.

  Appreciate the comments and points taken.

  Granted easier said than done.

  Cheers gs

• Is XIV relevant? Actually not in it's current form, the one-size fits all ethos quite frankly is bonkers! It's very attractive initially but you quickly come across problems and find it is not appropriate for all workloads. But...the concept of building an array out of commodity components and utilising Grid/RAIN type architectures is one that will probably grow and become more common.

  This is itself is not unique to XIV mind you and there are quite a few products around which do this; certainly in the digital media space, this is not an uncommon approach.

• Martin interesting points about the approach, trend of using scale out and scale up for future approaches.

  What I find also interesting is how much noise IBM and the XIV folks make about rain/grid/cluster approach with XIV and commodity components. Yet, how little noise is made about how DS8000 is based on off the shelf (granted IBM) components in the form of AIX plus pseries as platforms.

  Sure, they are IBM propritary vs. intel, however have you seen or heard of an XIV running on a Dell, HP, SuperMicro or other vendors data/storage server hardware platform yet? ;) Likewise, while IBM articulated the vision several years ago of how many nodes as an architecture DS8000 could scale to, forget if it was 8, 16 or higher, however many more than currently implemented.

  So concur, as a trend, direction more modular scale out solutions are becoming more common, some have been around for years if not longer in some propritary approach vs. others. Likewise, some of the traditional monolithic solutions are becoming much more modular looking more and more like their cluster/grid/rain counter parts.

  Thanks for the comments and discussion.

  Cheers gs