« I Don't Know Quite What To Think ... | Main | Do-It-Yourself Storage »

April 25, 2008

Comments

the storage anarchist

Seems that you and Dave Raffo are on the same wavelength today:

http://storage.blogs.techtarget.com/2008/04/25/primary-storage-still-means-one-vendor/

mgbrit

Yes, we've heard this many many times over the last 10 years. It absolutely makes sense not to mix your vendors if you're looking to simplify your business operations and improve utilization.

I would take this argument further and consider that simplifying across the app-OS-server-storage stack is also significant.

Why more companies don't buy their storage directly from their server vendors always surprises me.

Chuck Hollis

Why don't more people buy storage from their server vendors?

For probably the same reason they don't buy networks, databases and applications from their server vendors: they're usually not very good at it!

But you knew that, didn't you?

mgbrit

I'm not sure if I agree with your comment, Chuck.

Quite a few customers must think they are good at it, which is why IBM, HP and Dell are second, third and fourth in sales, with pure plays NetApps and HDS behind them. Yes, obviously EMC still maintains #1.

Customers are increasingly buying Procurve from HP, DB2 from IBM and as for applications, both HP and IBM have dozens to sell - and they do!

Chuck Hollis

So, if you want to go a few rounds on this, I'm up for it.

If you study the market share numbers for any amount of time, you'll notice that the vast majority of the server vendor's storage numbers are smaller arrays sold at the same time that servers are sold, usually to smaller organizations.

Conversely, when the organization is larger, or the applications are larger, the strong preference is towards storage specialists.

That's for storage. If you'd like to get into the market share numbers for databases, applications, etc. -- I'm sure I can dig those up, but we both know what I'd find, don't we?

Bottom line: server vendors seem to do best at servers, and little else.

tarheel

Unfortunately, smart engineers don’t make the decision because they would not just focus on capital acquisition like executive management does they would focus on TCO. We have several server vendors who constantly try to package the storage with the servers and propose the packages to management as a way to save capital but they refuse to talk about the TCO of the package they propose and its complicating effects on the storage environment. That is how we have ended up with so many vendor products on the floor. As I like to say, the server vendors sell their inferior storage to upper management because they cannot sell it to the technical guys.

Chuck Hollis

Hi Bobby -- good to hear from you.

You're right, an awful lot of storage gets sold that way -- more than most people think.

For example, if you look at the market share rankings, you'll notice that IBM, HP, Sun and a few other server vendors are usually ahead of a specialist like NetApp or HDS.

And you say to yourself "why is this?" because you *know* these server vendors aren't really committed to storage R+D, specialization, etc.

I mean, who in their right mind would buy HP storage to go with their IBM servers, or IBM storage to go with their Sun?

If anything, "strongly encouraging" customers to buy what's essentially proprietary storage to go with their servers is one of the best lock-ins left to the server vendors. If you have their storage, you're far less likely to go looking for a different flavor of server, right?

Thanks for writing!

Joerg Hallbauer

Chuck,

I think that block storage virtualization might be the sea change when it comes to this topic. I've bloged on this recently (http://joergsstorageblog.blogspot.com) so I won't repeat it all here, but I think that BSV might allow a company to get the best of both worlds. You can buy your storage arrays from whoever is cheapest per GB at the time without paying the "multi-vendor penalty".

Unfortunately, the same can’t be said for what most folks refer to as “file virtualization”. Global name spaces allow us to move things around transparently, and that’s a good thing. But they don’t address the people training, process, and procedure end of things like BSV does.

--joerg

Chuck Hollis

Hi Joerg

Your line of thinking is quite common among many people, but I disagree -- especially in demanding environments.

Here's why.

Consider any block storage virtualization device with arbitrary storage arrays behind them.

And now, let me ask three simple questions.

1 -- Who supports the end-to-end environment? The storage virtualization vendor? No, they didn't make the arrays. The underlying array vendors? No, they didn't make the storage virtualization device.

Claims to the contrary, it often ends up that the customer ends up shouldering more of the debug and support load -- there's no longer one throat to choke. Maybe that matters, maybe that doesn't.

And, our anecdotal history has confirmed this from many larger customers who've tried this in one part of their environment. And they're not gonna do this anymore.

The exception might be when the storage virtualization vendor has made specific investments to support an end-to-end multivendor environment, and is willing to write an SLA around it. EMC, incidentally, does this for Invista, our BSV.

2 -- What storage array features are available outside the virtualized domain?

Answer none: the BSV doesn't usually have any unique knowledge of array-specific replication, performance optimization, error-handling, etc.

You're limited to whatever functionality the BSV brings to the table -- almost none of the underlying storage goodness is easily leverageable.

They'd better be pretty dumb arrays indeed.

3 -- Are you really saving money and providing a better solution? Run the costs for buying dumb(er) arrays and BSV, as compared to a more traditional in-the-box solution.

We have -- many times -- and the numbers keep coming up the same. You end up with a more cost-effective, more functional and more reliable approach without all the additional complexity.

The exception might (might!) be a customer with a large number of "stranded assets" that are on the books at an unreasonably high value, and wants to pool or consolidate them to wring a bit more life out of them.

But even that's an artifact of unrealistic accounting procedures, and not technical reality.

Still, all that being said, there are a few vendors out there who can only hang their hat on their ability to do block storage virrtualization.

And neither of them seem to be doing too well in the marketplace, IMHO.

Cheers!

Amy

Nice post

The comments to this entry are closed.

Chuck Hollis


  • Chuck Hollis
    SVP, Oracle Converged Infrastructure Systems
    @chuckhollis

    Chuck now works for Oracle, and is now deeply embroiled in IT infrastructure.

    Previously, he was with VMware for 2 years, and EMC for 18 years before that, most of them great.

    He enjoys speaking to customer and industry audiences about a variety of technology topics, and -- of course -- enjoys blogging.

    Chuck lives in Vero Beach, FL with his wife and four dogs when he's not traveling. In his spare time, Chuck is working on his second career as an aging rock musician.

    Warning: do not ever buy him a drink when there is a piano nearby.

    Note: these are my personal views, and aren't reviewed or approved by my employer.
Enter your Email:
Preview | Powered by FeedBlitz

General Housekeeping

  • Frequency of Updates
    I try and write something new 1-2 times per week; less if I'm travelling, more if I'm in the office. Hopefully you'll find the frequency about right!
  • Comments and Feedback
    All courteous comments welcome. TypePad occasionally puts comments into the spam folder, but I'll fish them out. Thanks!