NetApp 50% Virtualization Guarantee Program

There has been a lot of storage vendor hand waving recently about green storage and storage efficiencies.  While everybody has a story to tell, what it boils down to is "How real are these benefits?"

Furthermore, how many times have you heard perspective customers question the  validity of various "Storage Efficiency vs the Competition" studies vendors provide, claiming the results are skewed in favor of vendor X simply because vendor X paid for the study? After all, It is a logical assumption and it is part of human nature to always question those who provide such reports. So, why should customers believe our Efficiency claims? Because we're ready to firmly stand behind them in a manner than no else in the industry has.

Today, at NetApp, we decided to put our money (and disks) where our mouth is. So rather than provide some concocted "Efficiency" Report invented behind the closed doors of our labs, we opted to announce a new program. The,

 

 

"50% Virtualization Guarantee Program"

This program is designed to show NetApp customers how to save 50% of their storage versus comparable competitor configurations that extend similar performance and protection levels by following documented NetApp Best Practices as published in various Technical Reports and the Program requirements.

In fact, we, at NetApp, are so confident in the savings customers will experience, that we have established a customer Program backing the validity of our claims. 

Program Details

You can see the details for yourself , but here are some of the highlights:

• Work with your NetApp sales team or NetApp partner to size the system you'll need
• Use any VMware supported protocols (FC, iSCSI, NFS)
• Set up aggregates (RAID protected storage) with 14+2 drives (or more) using RAID DP
• Place your common OS images (typically 10 or more) into NetApp volumes
• Run NetApp deduplication
• Repeat as often as you like for up to 6 months past your purchase date
• If you don't see the saving we've promised (and you've followed the best practices), we'll help you resolve the issue or provide you with the required capacity free of charge
• The Program is available Worldwide thru Direct NetApp and Channel Sales

Note: The program only applies to new systems purchased under this program, but if you already have NetApp storage system you should be able to get the same saving by following our Best practices.

Why did we do this?

There are a lot of bold marketing claims out there, and providing this program is way of showing customers that we believe our practices are meaningful, real and provide a tangible outcome.

However, I think the real reason we're doing this is because we believe space savings only address one aspect of the efficiencies NetApp provides in a VMware environment.

Once customers see the other advantages we have, in terms of performance efficient snapshots, rapid and efficient cloning and provisioning techniques, rapid data backup and recovery, and un-compromised data protection, they will realize a whole new way to manage their storage in a VMware environment.

Come for the space savings, and stay for the simplicity, efficiency and ease of use. Wouldn't you like to try it?

Disclaimer: The guarantee and the related Program is limited to the terms set forth in  www.netapp.com/guarantee, and is applicable only to prospective orders placed after the Program effective date and is dependent upon customer’s compliance with the terms and conditions set forth in this document and any of the instruction sets and specification set forth in the referenced documents. Netapp’s sole and exclusive liability and customer’s sole and exclusive remedy associated with the terms of the guarantee and related Program is the provision by NetApp of the additional storage capacity as set forth in the Program.

Technorati Tags: VMware,NetApp,Storage Efficiency

Comments

[Chuck Hollis]

Hey Nick, nothing from nothing, but this "guarantee" seems to have so many exclusions, caveats and limitations one has to wonder if it was intended to actually be of value to customers.

See my take on it here:

http://chucksblog.typepad.com/chucks_blog/2008/09/storage-shenani.html

[Nick Triantos]

Hi Chuck,

First of all first thanks for posting.

So, instead of going down the path of defending every single point raised, I will ask a simple question. What is EMC's Guarantee today?

You've raised the point of comparing vs RAID 10. we think that's relevant given that EMC's published document (h2197) titled "VMware ESX Server Using EMC Clariion Storage System Solutions Guide" there's the following statement recommending a complex mixure of RAID 5 and RAID 10 configuations:

“Virtual machines that are anticipated to have a write-intensive workload should use RAID10 protected devices on medium size, fast Fiber Channel drives… The log devices of databases should be on RAID10 protected devices… The virtual machines that generate high small block random I/O read workload, such as Microsoft Exchange, should be allocated RAID 10 protected volumes.”

http://www.vmware.com/vmtn/resources/540

Thanks

[Chuck Hollis]

Ummm, Nick, I think that's known as a "non sequiter".

The "NetApp Guarantee" specifically excludes high I/O and high write workloads. Also, there's a disclaimer that more spindles might be needed to meet performance expectations.

So I guess we're not really comparing the same thing, are we? And you're not seriously comparing RAID-DP as equivalent to RAID 10 in either performance or availability, right? Sure, RAID 6 has its uses out there, but that's not what the claim being made is.

C'mon, I know that there are real engineers over at NetApp. How did you let the marketing guys get away with this one?

-- Chuck

[Nick Triantos]

Chuck,
The Guarantee Program is intended for comparison with traditional systems using the same protection level (RAID-DP) which is the default NetApp's protection scheme.

As far as the performance aspect of RAID-DP with RAID10, although, performance is not part of the Guarantee Program as you pointed out, these questions were answered back on January 29th with our SPC1 audited results comparing RAID-DP vs RAID10. Although, we know how EMC now feels about the SPC1 benchmark, the fact of the matter is that EMC, along with other vendors, help had worked on it.

Having said that, going back to my original question, what is EMC's Guarantee today?

[Chuck Hollis]

But wait, Nick, you're not making sense here.

The presumption your making is that RAID-DP is both AS FAST and AS RELIABLE as RAID 10. SPC-1 monkey business aside, clearly your not claiming both, are you?

And, my guess is that you're enough of an engineer to understand what the SPC-1 really does, so I understand if you don't push that argument too far ...

Why didn't you make the guarantee against other RAID 6 implementations in the market? Or other thin provisioning implementations?

Oh, wait, then you'd have to make good in all sorts of cases ... I get it.

And, BTW, what's with the exceptionally long list of caveats, exceptions, and so on?

Can't include snap reserves.
Can't include virtual LUN expansion pools. Can't have a buffer for pre-dedupe data. Can't use a database.
Can't use Exchange.
Can't use data that's not easily compressible, like jpgs, ppts, XML, etc. Can't be high I/O rates that require more spindles.
Many similar VMs on the same flexvol.
Performance may require more spindles.
Have to buy an entirely new box to get the goodies.
Have to sign up to a full boat of NetApp pre and post services.

Yikes! With all of that, what CAN you use it for, and still be part of the "Guarantee Program"?

And, finally, to answer your question, I think I answered it in my post.

EMC is not exactly a fan of these sort of marketing stunts, so you probably won't see that sort of "guarantee" (in quotes for obvious reasons) anytime soon from us.

As a matter of fact, we do engage with customers routinely to drive out extra storage utilization in their environments, but it's an end-to-end professional service, and EMC doesn't get paid unless we deliver the promised capacity results AND maintain performance AND maintain availability AND don't break their existing IT operational processes.

I guess you could call that a guarantee of sorts. Kind of a different approach to customer success, wouldn't you agree?

Best regards --

Chuck

[Nick Triantos]

Chuck,

As I mentioned in the post, we're comparing against solutions providing similar data protection characteristics for VMware environments. Now, while RAID 6, provides similar data protection levels vs RAID-DP, typical RAID6 implementations offered by competitive solutions, don't even register in the radar versus those using RAID10 or RAID1 configs. Although the program provides a 50% guarantee, It is my personal belief, based on experience, the actuall % customers will see will probably be higher.

We stand behind the Program Guarantee and we believe that some customers will find it beneficial similar to EMC's belief on implementing Powerpath on ESX server. Others may not. Time will tell.

As far as the thin provisioning aspect of it, I've noticed Chad talking about and showing Virtual Provisioning for VDI deployments. In fact, I believe the VDI demos EMC runs use Virtual Provisioning. So, you really can be dismissive of it when you in fact are talking about it and showing it.

Which brings me to your comment about EMC not been a fan of such guarantees or "marketing stunts" as you called it.

If I recall correctly, EMC was for a while dismissive of Thin Provisioning. Until EMC introduced it and Joe Tucci called it "mandatory in the storage landscape" in April 2008 during an earnings call.

So this "not a fan of marketing stunts" statement sounds like a cope out to me and adheres to the EMC marketing standard of "if we can't do it or don't want to do it, lets diminish its value".

Cheers

[Chuck Hollis]

Hi Nick -- I don't think you're following the thread of the argument here at all.

I can't tell whether that's intentional obfustication, or whether you're just a bit confused.

Let's take your first claim: that somehow RAID-DP is a direct replacement for RAID 10 in terms of performance and availability.

Here's a workload running on 8 spindles. Add 2 and we now have RAID 6 (or DP, if you prefer) for a total of 10 spindles supporting the workload.

Now compare that with RAID 10 running the same workload on 16 spindles.

And you claim the performance will be equivalent? You must have a pretty low opinion of people's technical acuity, or magic juju disk drives that no one else has.

Wait, that would be flash drives ... :-)

Homegrown SPC results not withstanding, simple logic doesn't support your position, does it?

Ditto for availability, right?

Now, if you go back and look at NetApp's announcement, you decided to compare against "traditional storage" which often supports both RAID-6 and virtual (or thin) provisioning.

I'm not dismissing either (as you mistakenly insinuate -- where'd you come up with that?), they're now a standard feature of the storage landscape for specific use cases.

Hello, what's the NetApp advantage here?

No, all of it rests on dedupe for primary storage, doesn't it?

And, if you think about it for a minute, there's nothing special about a VMware environment, is there?

Here are 50 applications before virtualization. Here are 50 applications after virtualization.

Look, everybody, almost all the data is the same!

So why the mandate that this only applies to virtualized environments?

I am absolutely sure that NetApp stands behind its "guarantee".

There are enough caveats, exceptions, heretofores, prerequisites and process that I can't imagine anyone, ever, seriously trying to take you up on your offer.

There's absolutely nothing for you to "stand up" to, is there?

And I'm just waiting for the claim a few months from now that "we've never had to make good on our guarantee" simply because nobody could find any use case that didn't run afoul of your multiple exceptions.

Even more randomness in the rest of your comment. Nick, it's very hard to follow your line of thinking ...

As an example, how can you equate your latest "guarantee" with EMC making PowerPath available native on ESX? Did I miss some subtle yet important connection here?

Finally, EMC's approach to this (e.g. services-led engagement with defined outcomes and no pay if no results) works on anyone's storage of any vintage, new or otherwise.

It's not positioned as a sales promotional gimmick. Big difference between the two companies, right?

That's what I mean by "marketing stunt". I.e. grab a lot of attention, but don't really do anything useful for customers.

Look, I know you have to defend your company's announcements, no matter how boneheaded.

But please, don't assume that people aren't smart enough to figure out what's going on here.

-- Chuck

[Nick Triantos]

Chuck,

Guarantee program aside, we've proven that RAID-DP can perform better than a Clarion configured with RAID 10 against an industry developed benchmark using a cache hostile workload. As far as your statement about "homegrown" SPC results, let me remind you that the SPC1 results were audited and that EMC had 45 days to respond.

To qualify my PP statement better...per a conversation I had with Chad in a previous blog, EMC believes that porting PP to ESX will benefit customers and that it will ultimately be up to the customer to decide whether to chose native multipathing or PP. This is a fair statement.

In the same manner, we believe that our Guarantee Program will benefit customers and it is up to the customer to decide whether they want to participate. This is also a fair statement.

As far as your RAID6 example.

Our default and best practices for RAID-DP configs state 14+2 Raid Groups with a 12.5% overhead. Thus we are writting and reading to and from 14 Data drives vs 16 on RAID 10.

In typical fashion between the 2 companies, we agree to disagree on many things.
There's nothing wrong with that, as long as we don't become disagreeable...I believe using comments such as "bonehead" is unwarranted and offensive.

Cheers

[Chuck Hollis]

Thanks for staying with me here, Nick.

OK, so you configure 14+2 with RAID-DP. The same amount of capacity using RAID 10 would be, ahem, 28 spindles, right?

Are you arguing that 16 spindles are just as fast as 28 spindles? Or just as reliable, for that matter?

Think carefully, amigo -- your response will say a lot about the SPC and NetApp's use of it :-)

Your "choice" argument leaves me somewhat confused. Customers can choose to do most anything. I must be missing something very important here.

Just to set the record straight, only SPC members can challenge SPC results.

EMC is not a member of the SPC, nor did we collectively think any of this was worth the trouble, since it was obviously -- well -- just another marketing stunt.

Good luck with this one, guys!

[Nick Triantos]

Hi Chuck,

A couple points on your comments

1) RAID-DP or RAID6 is more reliable than RAID10. If you don't believe us you ought to believe EMC's own white papers such as document H2891 titled "EMC Clariion RAID 6 technology: A detailed review" published on July 2007 which states the following on pg.10 under the heading "Reliability":

"RAID 5 and RAID 10 have good reliability while RAID 6 has great reliability"

2) As I mentioned before we use dy default RAID-DP raid groups of 14+2. You are providing an example of a 28 RAID 10 config? As you should know, the min-max number of drives in a RAID 10 config on a Clariion starts from 2+2 up to 8+8 This is stated in the Clariions publicly available docs. In fact, EMC's recommended RAID widths for RAID10 is 4+4 and various examples of that are given in EMC's own FLARE 26 best practices guides. So I have no idea as to the basis for your example when it can't be done on the Clariion and if it could, it's definately not supported.

So, assuming, you were to lay down 8 blocks on an 8+8 RAID 10 config, given that you can't do 28, you'd require 16 writes. In my case, based on how WAFL works (2 adjacent RAID stripes, 5 disk writes and one more for RAID-DP. In turn you'd be able to read from 16 drives vs my 14.

3) Marketing Stunt - Interested you mentioned that again because in my previous reply I forgot to mention EMC's giving away a car at VMworld. Now, that;s a marketing stunt. However, I've decided not to harp on it anymore than I have, because after all it is Vegas, and what happens in vegas stays in vegas. I would say though that EMC's marketing stunt certainly fitted the Vegas slogan.

Cheers