Tough Decisions For IBM DS8000 Users?

Now, please don't take this the wrong way. 

I'm not picking on IBM here.

But I think that certain customers that are using DS8000s are facing a hard choice -- and it's coming up sooner that later ...

What's This All About?

The DS8000 is IBM's "high end" storage array.  It's been out for quite a while.

Note: I put "high end" in quotes, simply because -- architecturally speaking -- it's a dual controller design with more in common with mid-tier architectures rather than multi-controller, dedicated cache designs such as DMX or HDS's USP.  But I digress ...

Simply put: there may be no "next" DS8000. 

And -- right now -- a lot of those boxes on customers' floors are looking very long in the tooth.

IBM hasn't been very clear -- either officially or unofficially -- about the future of the DS8000.  And there's a shiny, new toy -- XiV -- that looks to be the new heir apparent.

If you're a customer, you're caught in the gap -- you don't want to put more money into your DS8000s, but the XiV box isn't ready for prime time yet, and won't be for a while.

What are you gonna do?

Why So Many Of Us Are Sounding The Death Knell For The DS8000

The orginal architecture -- introduced many years ago -- was a dual-controller design based on the PowerPC architecture.  Although the "theoretical specs" sounded quite impressive at the time, very little of it came to pass: multi-controllers, LPARS that did useful work, bigger caches, more drives, etc. -- none of it really happened.

None of the investment materialized to make the potential a reality.

Sure, I could pick off the DS8000 feature-by-feature, but that's not the point.  What matters is that it looks like IBM has essentially given up, and moved on.

One thing that most people in the industry have noticed is that IBM has fallen off the "upgrade train" for the DS8000 for quite a while, and hasn't shown signs of getting back on the track anytime soon. 

As an example, during this multi-year period, EMC has predictably marched from DMX 1 to DMX 2 to DMX 3 and most recently DMX 4. Every 6-9 months, you either get a major update, or an entirely upgraded design.  Not to mention a boatload of cool new features at predictable intervals.

Not too many customers lose sleep wondering where the DMX is going, and what's going to happen to their investment.

HDS hasn't been as aggressive, but at least they've stayed in the game.  IBM?  We wonder if they're even playing anymore -- or have decided to move their investments to another platform.

Sure, you could criticize them a bit on dated processor hardware, performance, capacity, connectivity, et. al. -- but where the differences are most noticable are the newer, money-and-power saving features showing up in the high end: thin provisioning, terabyte drives, enterprise flash -- not a peep that they're even thinking about it for the DS8000.

To their credit, we've seen a few software updates over the last few years, but not the sort of sustained investment that would indicate that the DS8000 is a flagship product for IBM, and is going to be around for a while.

And Then There's XiV ...

The story just wouldn't be complete with a bit of commentary around what appears to be going on with XiV.  Moshe Yanai appears to have done a good job of rounding up most of IBM's storage software developers for things like PPRC, FlashCopy, etc. 

If this is the same Moshe I worked with years ago, I can vouch for the fact that these storage software engineers aren't working on anything other than the XiV box.

Moshe has also assembled a dedicated sales and marketing force flying the XiV banner (looks to be outside the IBM mainstream -- intentionally), and we hear from a few customers that the Xiv team is quietly positioning the XiV box as the DS8000 replacement. 

Now, I don't know whether that's Official IBM Policy, but that's what tends to happen when you recruit a few aggressive ex-EMC sales guys and hand them a new product ...

Why This Is A Major Pain To DS8000 Customers

Many of the DS8000s on shop floors around the world are getting old -- really old. 

They're not as efficient or as cost effective as newer designs by a growing margin.  You might think that maintenance costs would be an issue, but I bet IBM will find a way to make that go away for these customers.

What IBM can't make disappear is the loss of performance, floor space, power efficiency, newer functionality, connectivity, etc. that modern enterprise arrays offer.  A lot has happened in this industy in the last 3-5 years, and IBM hasn't kept the DS8000 up with most of it.

I don't think IBM intends to write customers a check for those financial impacts ...

Meanwhile, the XiV is a new box with a new design.  It's going to take time to make it reasonably compatible from a functionality, stability and performance perspective with the current DS8000. 

As in "a few years" time. 

There's mainframe connectivity to add, a boatload of proprietary IBM storage array software to move over, not to mention an extended "baking in" time before it can be assumed that it's a reasonable substitute for the DS8000 in production environments. 

All due credit to Moshe and his team, but there are no shortcuts here, folks.

One could speculate as to whether IBM will be able to engineer a reasonable substitute for the DS8000 from the XiV starting point (long list of architectural reasons here), but -- for this discussion -- I'll give them the the benefit of the doubt.

That leaves a multi-year gap for DS8000 customers between "your existing array is dead end" and "here's the new replacement you've been looking for".

Not good, if you're using these to run your business.

So, What Should DS8000 Customers Do?

The first thing I'd do? 

Please don't take my word for it -- ask IBM for the straight scoop on what's up with the DS8000 and XiV.  If you're running your business on this box, you should know.

You may have to ask a few people with differing views: your IBM rep, the XiV rep, or someone from IBM management.  We've heard that getting a consistent story with documented roadmaps is a bit difficult, but at least you should try.

If you come to the same conclusions that many of us have, you may decide it's time to ride a different horse for your enterprise storage.

And we'd love to have the opportunity to earn your business ;-)

Comments

[mgbrit]

Chuck,
Good points and a nice summary of what a lot of people who follow storage in the industry have been discussing over the last 18 months.

So who has traditionally bought the DS8000 ?

I get to go in dozens of data centers a year and you see a lot of IBM storage in IBM shops or IBM storage attached to IBM gear in co-lo's, but you rarely see IBM storage outside of IBM shops or attached to non-IBM gear.

For many IBM customers, not a lot may have changed in the last 3 years and although new functionality may not have appeared on the product line, customers with a mainframe and an IBM Global Services or Sungard DR contract may not require new functionality.
When I come across a strong IBM shop, it tends to be in a regulated industry such as insurance, banking, or healthcare. If the regulations (state and/or federal) have not brought about a business change which in turn forces an IT change, the company largely continues to operate under similar SLAs with similar needs. Obviously this is a simplistic look at businesses, but the observation is a valid one.

IBM knows how to make money and had proven this in the Gerstner years by focusing on high-margin software and services. Even if IBM's hardware (desktops, servers, storage) is not best in class, as long as it is good enough, customers will buy it when they see value in other aspects of IBM's value proposition.

Moreover, EMC is focused on very specific areas of the data center. If a customer is buying a multi-million mainframe from IBM, buying millions of dollars of CA and other software to run on it as well as other application software, the storage may be a rounding error.
It's hard to compete for the business when IBM is bundling the lot.

[Chuck Hollis]

All good, insightful points, mgbrit -- thanks!

[Al Sharpton]

I have heard that the DS4000 is also approaching EOL. Hallelujah!

[Juan Jose Palacios]

DS4000 approaching EOL?

Please, I need to confirm this point asap.
Any reliable reference to get info from?

Will search for it right now.

Juan

[Chuck Hollis]

Sorry, I don't have any information on that one ... best of luck!

[David Hartman]

Hi Chuck, neither the DS8000 nor the the DS4000 are approaching EOL. The DS8000, despite being almost four years old, still offers attractive features, performance, and reliability for System z & Open workloads. There are also many feature & enhancement work items planned on the official product roadmap. Your article lists some salient points, but paints an overly pessimistic picture that does not exist for DS8000 customers that I am aware of.

[Jack Meov]

Chuck:

Get your facts straight. XIV is NOT replacing the DS8000. As usual, you are blogging about nothing...!

[Chuck Hollis]

Hi Jack

You're responding to a post that's a little over a year old. Are you aware of that?

-- Chuck