Why settle for bad Vanilla?

by Tom_Joyce on 04-16-2011 08:22 AM - last edited on 04-16-2011 03:50 PM

At my house we frequently get one of those half gallon containers of ice cream that has vanilla, chocolate and strawberry in the same box.  They call this "neapolitan", which sounds more sophisticated than it actually is. 
 
Typically it looks something like this:
 
 Neapolitan.jpg
 
At the end of the week, when dad finally gets a chance to hit the refrigerator, typically what I find is that most of the ice cream is gone.  All that is left is a ragged box with no strawberry and no chocolate.  Just bad vanilla.  The thing about a box of supermarket neapolitan is that it really isn't very good ice cream.  The attraction is that it's an "all-in-one" solution.  My experience is that the strawberry is usually pretty good, the chocolate is just OK and the vanilla is pretty bad. 
 
Now I have to confess, that typically I eat the leftover vanilla anyway.  Why?  First, my standards are not particularly high and second, because I paid for it.  But I hate settling for bad vanilla.  Especially when what I really wanted was Heathbar Crunch. 
 
Changing gears
 
One of the most successful marketing messages in storage over the last decade was the idea of "unified storage".  Simply put, a unified storage solution is one that brings together NAS, Fibre Channel SAN and iSCSI in a single box. NetApp coined the phrase "unified storage".  Their approach was to extend the NetApp filer’s NAS capability to also support Fibre Channel and iSCSI with their WAFL filesystem, with a common disk pool, unified packaging, management, and data protection.  They had a fair amount of success with this, especially in scenarios where a customer was primarily interested in NAS but might want to do some SAN storage too, now or in the future. 
 
This option to do SAN on the same box with the same tools was attractive to smaller companies with limited resources.  But in my experience, most of the customers that have implemented these systems are not using the Fibre Channel SAN capability.  They really just use the NAS.  This is because the Fibre Channel SAN capability on these systems is bad vanilla.
 
As I have said before, it is very difficult to optimize for everything.  It is very difficult to modify a NAS platform so that it can be an enterprise-class SAN platform too.  Significant compromises are required and also a lot of patchwork, like background tasks to do garbage collection in the filesystem to manage fragmentation and various other inefficiencies.  I think if you go out and ask ten storage experts the question, "Which company delivers the best Fibre Channel SAN capability?", the number of them that will answer "NetApp" is zero.  This is because it is not the best.  It is a set of compromises. 
 
For the last few years, NetApp has been working to expand their unified storage platform to include not just traditional NAS but also scale-out clustered file storage.  They acquired technology to develop this quite a few years ago and have struggled to bring out this capability, called GX, in a fully unified fashion.  My guess is they have found that too many compromises are required to bang all this stuff together.  In fact, NetApp's recent acquisition of Engenio for low-end SAN and DAS is a tacit admission that one architecture won't get them where they want to go.
 
Here's the deal
 
There are two pearls of wisdom in the concept of unified storage.  The first is that storage management across multiple protocols and platforms is too complex today and customers hate it.  As I have said before, the number one thing customers ask us for is to make storage simpler.  This is the thing NetApp got right, by providing one user interface and a single set of tools.  The second key idea is that pooling back-end storage capacity across SAN and NAS is attractive because it can provide better control over utilization and costs.  In unified storage implementations, the value of this has not been fully realized because of the inefficiency, fragmentation and reserves required but the idea is a good one.
 
So what is the HP answer?  We want to provide simplified converged management capabilities and converged infrastucture that is highly efficient, without compromising performance or efficiency.  You pick the flavors you want (FC, iSCSI, NAS, DAS, scale-up, scale-out...) and we will deliver converged management that makes things simple, and converged infrastructure that drives back-end efficiencies.  One advantage of this approach is that, unlike "unified", our "converged" extends beyond storage.  You get converged management with HP CloudSystem Matrix, which provisions storage, servers and networking in one go, and is very simple to use.
 
Here is what Converged Storage looks like:
 

Ice Cream Cones.jpg
 
Why settle for bad vanilla?

We encourage you to share your comments on this post. Comments are moderated and will be reviewed and posted as promptly as possible during regular business hours.

To ensure your comment is published, please follow our community guidelines.

Comments
by Dimitris Krekoukias(anon) on 04-18-2011 10:36 PM

Hi all, D from NetApp here.

 

Usually I ignore such posts but here's the short answer that I sincerely hope you will publish:

 

Some of the largest FC/iSCSI implementations in the world are running on NetApp, including some of the largest (if not the largest) DBs and email systems, for instance. 

 

Recently we were benchmarked in a PoC against an 8-node scale-out dedicated SAN architecture (guess who) and our 6080 box (not the newest or fastest) had 3.5 times less latency. Using the customer's DBs, not our own data.

 

NetApp is already huge and growing, we have nothing to prove any longer.

 

An article like this is almost like saying that HP has challenges achieving unification, therefore everyone must have the same challenges.

 

It's also insulting to the countless customers of NetApp doing block storage and achieving amazing results, and does you no favors.

 

Maybe NetApp got big exactly because we licked the unification problem :smileyhappy:

 

D

by Tom_Joyce on 04-27-2011 04:42 AM

Dimitris, thanks for taking the time. We are happy to see Netapp people on the site, and we love the debate.  I certainly agree that Netapp has been successful.  I have always admired them as a marketing company. They have changed their business from file server consolidation, to backup-to-disk, to"unified storage". We think the next wave of changes in the market are now afoot.  The next wave is about IT as a Service, scale-out, cloud, and converged infrastructure. It will be interesting to see how Netapp responds. Markets change, and so we all have something to prove, every day.

 

You're right, Netapp has gotten big. Your storage business is about the size of ours. But HP is really big. When your world is storage only, putting three storage protocols in one midrange box makes make it seem like you have "licked the problem of unified unified".  HP's world extends well beyond midrange storage, and as a result, our view of what is required for convergence is much more expansive.

Post a Comment
Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.

Find HP in Social Media

Facebook Twitter YouTube SlideShare Flickr
About the Author
Labels