An Interesting Experiment By IBM

Saw IBM's announcement of a baby SVC the other day.  TonyP was even offering it up in Hot Pink.

And I thought to myself -- this should be really interesting to see whether they have any success with it or not.

Why?

Because it's positioned right at the fulcrum point of one of the most interesting debates around this particular topic.

The Debate

One of the most persistent debates around the entire controversial topic of storage virtualization is "how do you want to build your storage array?"

Do you want to buy separate components (virtualization device and JBOD-ish storage bricks), or would you like an integrated approach, i.e. all in one box?

Now, if someone's got significant amount of stranded storage assets, due to the arcanary of corporate finance or some other unfortunate situation, I can see a bit of merit in recycling and pooling your old storage with some sort of virtualization thingie.

Not exactly a strategic viewpoint, but we all have to do what we have to do ...

But what I find interesting in this particular announcement is that it's positioned exactly where customers will have a very interesting choice between simplicity and complexity.

The Details

IBM states that the baby SVC is just like its big brother, except it appears to be limited to 60 drives.  If you assume that people will eschew using multiples (complexity, cost, etc.) that's an interesting segment of the market.

Lots of good 60-drive arrays out there from EMC and others.  One example is the hot-selling AX4, which starts at around $8K list for 3TB iSCSI, and grows to 60 drives. 

Such line-expansion downward is a typical move for any product manger, but you've always got to ask the question -- is the market substantially different in the new segment?

Market Requirements In This Segment

For example, I'd consider iSCSI support mandatory in this segment -- but no iSCSI support on SVC, which adds more than a bit of cost and complexity to the customer equation.  Having NAS access is a very useful form of pooling (hence the NX4) as well.

And then there's the simplicity angle as well.  For example, we had an independent firm go test the simplicity and usability of the AX4 and share the results here.  I just can't see how IBM's approach breaking things into two independently configured and managed domains (virtualization device and external storage) is really going to help SVC in this part of the market.

Finally, there's the positioning angle, which is especially important with the channel -- what product do I sell where?  What's the advantages and/or differences between the two?

Now, looking at things from IBM's perspective, it's pretty clear what you do when considering larger environments -- sell the big SVC, since the DS8000 is way behind the curve, and XIV isn't quite here yet.

But in smaller environments?  Not only does IBM have decent arrays that they get from LSI, but they've got their NetApp-based offerings as well -- all of which create an awful positioning mess from my point of view.

An Interesting Experiment?

That's why I think this one will be interesting to watch.  My belief system is that different market segments have very different needs.  And what IBM has essentially done here is transfer a belief system that works decently well in one segment to an entirely different segment.

What do you think?

Comments

[marc farley]

While the physical configurations of SVC installs are more complicated than installing systems, I suspect the licensing end of things will end up causing the most difficult problems for customers.

Other than that, I think the pink was a nice touch. Do they come in plushie wrappers?

[Chuck Hollis]

I thought the Hot Pink was a bold move, too.

Me? I'm more of a Fierce Green kind of guy ...

[Barry Whyte]

LOL @ the pink comments.

We've had many many requests for a lower entry price to an SVC world, and we've listened. While something like AX4 does as you say provide that entry level of attachment, it doesn't provide the same enterprise level functionality, nor the ability to expand as your business does. You need to buy another one, or a different one, and then have islands of storage.

With SVC EE you can consolidate those islands, so maybe you have an EMC AX4 today, and you think you need another one and 3 months ago it seemed like one was enough. The ability to pool and consolidate, and start to do more and more backups, and maybe get Thin Provisioning an no additional charge to get more out of what you have...

All with the knowledge that if you have a conversion path to larger clusters and the bigger brother SVC in the future without having to rip it all out and start again.

Of course you can always order the lower cost node hardware with the "classic" SVC license.

[Chuck Hollis]

Thanks, but I think you took it off-topic a bit.

For example, cost comparisons between pre-integrated or assemble yourself. Or comparisons in simplicity and ease of use between the two approaches.

Or the lack of iSCSI in this market. Or the positioning question.

I guess during a political debate, if you don't want to answer, you change the question.

-- Cheers!

[Barry Whyte]

Chuck, I'm no politician, thats for sure. I wasn't intending to be off-topic. I was responding to your question about why would you have two boxes when you could have one. Two allows you to grow as I suggested above.