Myth Busting: Storage Guarantees

Vaughn,
I’m generally a fan of NetApp and its storage technologies- single file flexclone and dedupe on vmware on NFS is really great stuff.
I have to disagree here though. The EMC guarantee is more powerful because the lack of caveats. The fine print on the 50% savings is compared with RAID10 and requires NetApp PS.
First problem is most people run vms on RAID5 or 6, not 10. Secondly, NetApp generally has their partners do the PS. This basically invalidates 99% of the cases where the guarantee applies.

Hi Vaughn,
So “If it sounds too good to be true, it probably is” applies to EMC, does it not also apply to NetApp’s guarentee?
Over my 15+ years in the industry, I’ve used just about every storage platform out there and currently don’t use either NetApp or EMC. However, I pay close attention to what both EMC and NetApp are doing as I feel these vendors are leading the pack in terms of virtualization. NO storage solution is perfect and both EMC and NetApp have strengths over the other. I’ve become increasingly dissapointed that you continue to go negative against a competitor versus what makes NetApp a good product, which it is.
Look at your information above, it’s extremely misleading. For one, why are you comparing SAN technology to NAS? Also, wouldn’t you see the same results as your vSeries if you added an EMC gateway?
All told, I enjoy your blog very much; I just wish you focus on us the customer instead of trying to bash the competition. Your technology is too good to treat it otherwise.

Hey Vaughnn
Great blog here. I hope customers find a sense of reality after reading through this as there is a significant amount of marketing spin from all vendors. While i enjoyed reading the entire blog, the table is something that customers will understand and be able to use when evaluating not just Storage Guarantees but leveraging ways to really drive storage efficiency in their environments.
thanks for the great read

@Just A Storage Guy – I agree with you that the NetApp ‘guarantee’ is too restrictive and overly conservative. If I had my way we would never have included the RAID-10 component as a point of comparison in the guarantee, but alas that decision was not mine. With that said the average customer results ranging between 50% & 70% are common place, and have been shared by many in many forums (blogs, tweets, docs, etc.) What is interesting is customers don’t calculate storage savings based on RADI levels. They share their results based on usable data.
@Pete Smith & @Paul – Thanks for the kind words and the critique. I appreciate the contrast in your opinions regarding sharing information which helps potential customers to make an educated and informed decision. Storage arrays from EMC & NetApp aren’t exactly inexpensive.

If you are interested in real customers comments on their storage savings from Data Deduplication without including RAID-DP savings check out this post:
http://blogs.netapp.com/virtualstorageguy/2009/01/deduplication-g.html

Hi Vaughn — Chuck from EMC here.
You’ve offered up a fine example of what we call “chartsmithing” — the skillful crafting of a chart engineered to greatly mislead.
And you guys accuse me of propoganda 🙂
I think that you and the rest of the NetApp crew are missing some key points:
* the EMC 20% guarantee is in addition to any offer from NetApp or any other similar vendor. All the customer has to do is put the configs side by side.
* their are no caveats or exclusions regarding workloads, use cases, etc.
* we don’t rely on words like “probably” and “typically” — it’s a guarantee.
* we’re not insisting that people do really silly things, like start on RAID 1 and end up on RAID 6 to get the savings.
* there’s no PS required for the guarantee, nor are their weeks of training required either.
* the process is simple and straightforward.
So far, results have been pretty good.
The people who’ve compared have mostly found 20% or greater efficiency comparing EMC and NetApp configs side-by-side. This is true for small and large configs alike.
And, in the very few cases where we might be wrong, we do make good.
Customers win either way — and that’s the point!
— Chuck

@Chuck – thanks for the comments.
I do take offense to two of your points. Can you please elaborate one the following items?
1. How can EMC provide greater storage savings when EMC arrays offer significantly less storage savings technologies?
2. Which, if any, items in the chart are incorrect?
Chuck let’s be real… Are the slurs of ‘chartsmithing’ & ‘propaganda’ the depth of your ability to articulate EMC’s capabilities? Can you provide information in order to educate customers and those who are technically inclined as to EMC’s strengths around providing storage savings?
Please return and let’s continue the dialog.

/popcorn

Hi Vaughn
First, I am concerned that you considered my previous comment an “attack” on Twitter.
Seriously?
Now, back to the topic at hand …
Feel free to continue (a) make long lists of storage efficiency technologies, and (b) claim the superiority of them.
We could go toe-to-toe on all of them, but it wouldn’t really matter, would it?
All that really matters is the net result to the customer, and the ability of the vendor to stand behind those results.
Right now, the gauntlet is down in the marketplace: EMC is 20% more efficient. Guaranteed. No caveats, disclaimers or multi-page “feature charts” required 🙂
State your requirements. Get a config and a quote from both vendors, using their published guides. And compare.
It’s just that easy. See for yourself!
http://www.emc.com/products/unified-storage-guarantee/index.htm
It’s not surprising that your customers and prospects are increasingly taking us up on the offer.
Great unified storage products from the recognized industry leader, best-in-class VMware integration, world-class support — and at a great price!
It’s a bottom-line discussion.
As far as incorrectness, I’ll turn it back over to you: which items do you believe attempt to intentionally mislead the reader?
You know and I both know the answer …
— Chuck

Vaughn,
My chart would be a little different, when it comes to space savings I’d highlight the IOPS/spindle advantage we have, especially for VDI. Celerra’s performance out of the box is … well … pretty average, especially measured on an IOPS/spindle basis. For performance bound workloads. Reducing disk spindles and costs is more than just about capacity efficiency technologies, it’s also about performance efficiency technologies. To be entirely fair, I’d also change the compression feature from a “tick” to an 8.01 feature, but I’d add in the relative inefficiencies of EMC’s replication solutions such as recoverpoint when it comes to preserving dedup/compression benefits (or anything much to do with IOPS or capacity efficiency for that matter).
given that Chuck wont (cant?) come up with any details, of how EMC justifis their efficiency benefits, it comes down to this.
1. FC and iSCSI workloads dont need filesystem overheads – which is true, but only if you give up features like compression and advanced integration with Vmware
2. their SATA solutions dont need the Block Checksum overheads in OnTAP – which is also true except that their slipmask method imposes nasty performance problems for a number of workloads. In general to achieve the same balance of performance and avaialbility they have to use RAID-10 for SATA, which relegates their SATA for data at rest.
After that, there’s not a lot in their kit bag. Their fixed snapshot reserves leads to overallocation of snapshot space, the performance impact of iSCSI LUNs on celerra (or anything on celerra for that matter), prevents them from being reccomended, the same goes for thin provisioning for high performance workloads on Clariion.
The NetApp guarantee was devised to highlight the technical superiority of our solution set for virtualised workloads (especially those which were typically virtualised at the time of the guarantee).
If the EMC offer is genuinely “no matter what NetApp offers, we’ll do better” then this is not an offer based on superiority of technology, but one of a willingess to throw free storage at a customer to win a deal.
Overall, to me it smacks of desperation rather than a belief in the superiority of their solution.

“We could go toe-to-toe on all of them, but it wouldn’t really matter, would it?”
Actually, it would help potential customers.
The chart Vaughn showed was skewed toward NetApp, but let’s go toe-to-toe on features. It’s the only way potential customers are going to be able to cut through the marketing and get down to which solution benefits them the most.

I find it amusing that @Chuck accuses you of “chartsmithing” (who makes these words up??), but provides no detail as to what was ‘smithed’, instead attempting to hand wave the whole thing off with completely unprovable assertions!
@Chuck, rather than acting as if the whole world should ‘pay no attention to the man behind the curtain’, how about a substantive conversation, disproving any points on Vaughn’s chart you find annoying?
That kind of approach would promote transparency.
The simple truth is this – there is no single silver bullet that bring about storage efficiency. But it is a variety of pieces and parts working together in concert that bring it to life. In order to have this happen, you need real live Unified Storage, from stem to stern in your data center.
This model has helped Cisco rule the Network, and is driving NetApp toward the same destination. And it would not matter one wit – if customers did not benefit first.
@Brandon is correct. let customers decide based on facts and transparency.

@Chuck
In reading your replies I see claims from a industry giant attempting to defend its market share in the minds of its customers. What is missing in your replies is an explanation of how EMC with can actually deliver more storage savings than NetApp when they have fewer technical capabilities.
Are you expecting the world to believe EMC file level dedupe provides greater storage savings than block (or sub-file) level dedupe?
You once said that you would leave EMC if they ever pulled a marketing gimmick around storage efficiencies. Do you recall your words? I’d like to ask you to provide technical experts to discuss the EMC capabilities, or maybe it’s time you live up to your word.
I’m truly sorry to have to have made that last statement, but alas you drew the line in the sand declaring for truth and transparency, not I.
Chuck, you cannot have it both ways.

@John
Thank you for the comments, you are a man among men in the storage industry.
You make good points, the general availability of file system compression with NetAp is data ONTAP 8.0.1. But it is in use customers today who had early access to the code with 7.3.
EMC’s move to use RecoverPoint as their sole ‘unified’ replication tool is another excellent point. RP requires additional hosts, and has no storage savings intelligence, it is a pure block level replication tool that cannot leverage storage intelligence in the array.
I feel if I edit the chart to position RP, the EMC field will cry that they have alternative means even if these means are being phased out. So is it worth the edit?

@Brandon & @Mike
Great comments, thanks for sharing.
Customers spend hundreds of thousands to millions of dollars on a single enterprise class storage array. They deserve to understand the details of what they are purchasing relevant to their dataset, application, and business goals.
Virtualization allows a customer to drop a storage partner without any downtime and incumbent storage vendors are no longer protected by applications which are just too difficult to migrate.
Customers have power and vendors better deliver on their capabilities and claims or they will go extinct.

It’s pretty obvious Chuck (and the rest of EMC) can’t come up with ANY specific examples because this “marketing stunt” (to use EMC’s taxonomy) fails under the harsh light of reality.
I find it amusing that this company which defends the need for a complex & incompatible product portfolio due to the supposedly variable and complex requirements of their customers – now can’t come up with any of those said requirements to defend this thinly disguised discount program.
Mike Richarson dissects it well over here:
http://blogs.netapp.com/dropzone/2010/05/playing-to-lose-hoping-to-win-emcs-latest-guarantee-part-3.html
My question – once EMC tricks their victims into proving this out and then inevitably provides the additional storage to compensate – who pays for the ongoing OpEx costs of additional power, cooling and floor tile space?

The only Chuck that can have it both ways is Chuck Norris .. when it comes to storage .. you guarantee him that it will be 20% more efficient.

So how about applying some maths to the buying process – and you might spot the confidence trick.
Nice unsuspecting customer guy goes out to buy say 100TB of usable storage
NetApp guarantees to do it with 50% less – OR WILL MAKE UP WHATEVER IT TAKES to match so NetApp promises to deliver 50TB usable and makes 100TB guarantee.
EMC jumps in and claims to be 20% more efficient – than what ? Not the 100TB original as the customer is not going to buy 80TB from EMC when they only need 50TB from NetApp. So they say it is based on the NetApp 50TB.
So they offer 50TB EMC which will be 20% more efficient? huh? lets just say OK for now, as IF IT IS NOT THEY WILL DELIVER 20% MORE! for free. wow!
Hmmmm.. so 20% of 50TB is 10TB so they will make up the difference to 60TB even if the customer needs 100TB ?
Well the customer has been conned in two ways:-
1. If EMC cannot deliver the efficiency then he is not protected to 100TB
2. If EMC cannot deliver the same efficiency as NetApp (50%) then they still get the sale and deliver 50-60TB which is 40-50% more efficient.
So it would appear that Vaughn’s suspicion is right, EMC is nervous about delivering the same level of efficiency saving, 40% seems more likely, and the rest?
Well it is a clever subterfuge to sell 50-60% as much storage at a discounted price but win the deal.
This looks like an entirely cynical STOP-LOSS MARKETING CAMPAIGN and this is where you have to recognise how brilliant Chuck is.
He got you believing it was better than your 100% guarantee (sell 50TB but guarantee you will provide up to the original 100TB requirement.
So have I got the sums and assumptions all wrong – too much coffee and paranoia?

Don’t fret about Chuck, most enterprise customers like us are sophisticated enough to recognize that his primary role at EMC is to spin and possibly coordinate spin from his cadre of fellow EMC bloggers (Chad’s group excluded). I’ll never forget when I first mentioned Chuck’s name to our account rep, he visibly winced. You really are wasting your time if you think he is going to go toe to toe with you on specific technical topics. Thankfully he doesn’t represent the norm at EMC or we wouldn’t spend millions of dollars a year on their products (almost entirely on DMX\V-Max). Surprisingly, I’ve never seen EMC really go after what I consider to be NetApp’s greatest weaknesses, support and services. In our experiance EMC, hands down, is the better company when it comes to quality of support and services resources.

Chuck is indeed marketing. But I absolutely do think it would be worthwhile to bring a couple of guys together to discuss and refute key advantages, and disadvantages of the products.
But hey, I’m just a customer. I’m sure the spin machines would rather spin FUD than discuss facts that might help customers.
I’d even be willing to host such a discussion, as a third party who doesn’t yet own a single product from either company.
Let’s set it up. Or am I the only one that would like to cut through the marketing?

Does Netapp really have NAS/SAN compression as stated in the chart?