NetApp's Prestige

On Monday of this week we reviewed EMC's botched attempt at a multi-vendor storage efficiency comparison and yesterday we debunked some lingering misconceptions around provisioning capacity for LUN's on a NetApp SAN.

It was probably not Chuck's intention, but he's actually helped bring incremental attention to NetApp's industry-leading storage efficiency via some interesting disclosures by EMC and Industry Experts along the way.

Time to get back to the movie analogies and wrap-up what we've learned as a result of it all

The Prestige (Spoiler Alert!)

Recalling Michael Caine's haunting narration (first two quotes on top), we previously reviewed Chuck's tricky Pledge and how the tables have Turned on him. Now it's time to amaze!  This is how the film describes the end of a successful magic trick:

But you wouldn't clap yet. Because making something disappear isn't enough; you have to bring it back. That's why every magic trick has a third act, the hardest part, the part we call "The Prestige"

Much like the series of gruesomely botched magic tricks in the movie itself, this is where Chuck and his competitive team failed.  Fortunately in Chuck's case, no pretty assistants or birds died, and no fingers were blown off, cut off, or crushed.

However as per the ill-advised dismissal of Nicola Tesla's science portrayed in the movie, what EMC (are they channeling Edison via their intimidation tactics?) fails to grasp is that they are up against advanced technology in the form of NetApp FAS arrays - which is truly innovative and REAL, not sideshow magic.

The Sacrifice

Paralleling the moral at the conclusion of the movie, EMC customers are also often forced to sacrifice.

• Choose RAID5 and sacrifice system availability
• Choose RAID1/0 and sacrifice half your usable capacity before other overheads
• Choose RAID6 or space-efficient snapshots and sacrifice performance
• Choose LUN clones and lose the ability to retain numerous online recovery points
• Choose to deploy sexy new features like "Virtual" (sic) aka "Thin" Provisioning - and wonder where the best-practices or performance sizing tools are?
• Choose to believe EMC marketing, and sacrifice all best-practice recommendations

NetApp engineering also understands need for sacrifice - by resisting the temptations of an instant-gratification product development path consisting of designing a different solution to every storage requirement.

Instead NetApp FAS engineering concentrates on the more difficult yet satisfying challenge of delivering a truly unified storage solution with unique attributes such as a storage marketing value prop (availability, performance, efficiently and simplicity) which is:

• Matched by actual best-in-class best-practice deployment recommendations for RAID-DP, Thin Provisioning, Snapshots / FlexClones and of course Pervasive End-End Deduplication, and 
• Matched by objective and independent benchmarks validating all of the value-add functionality in realistic configurations as opposed to (stripped-down features on hot-rod hardware on) competitors' "benchmark specials".

The Tactical Perspective

In one sense, I can't blame EMC for their myopic competitive stance.  It stems from the limits of their engineering ambition.  They've simply given up on innovating a modern RAID scheme (now recognized directly by Microsoft in their Exchange 2007 Storage Recommendations) that enables the best of RAID5 efficiency and RAID1/0 performance, with even better overall availability.  And they've also given up on Snapshots which don't sap performance while using pointer-based efficiency.

The Strategic Perspective

What external events motivated EMC to do this, how they timed their attack for maximum PR exposure and minimum competitive reaction time - and what they might do next are all probing topics I will expose in a subsequent blog.

The Root of EMC's Confusion

It's EMC's relatively new introduction to the capabilities of Thin Provisioning which probably has them stumped.  Unlike the disparate Celerra, CLARiiON or DMX platforms, NetApp's Universally Unified FAS architecture enables the advanced capability to Thin Provision BOTH data AND snapshot space.  FAS arrays can do so on the same volume with the same planning, deployment, management, protection / retention and troubleshooting characteristics across the board. Form megabytes to petabytes on the industry's most scalable and consistent storage product family.

In comparison with legacy storage arrays like the EMC CLARiiON, NetApp FAS arrays can be de-tuned purely for basic capacity and single-parity storage efficiency with little or no regard for near-CDP capable application-consistent recovery points (NetApp snapshots) or advanced RAID ...

BUT with RAID-DP & Snapshot Thin Provisioning Enabled, NetApp FAS arrays have proven they retain that same industry-leading efficiency while simultaneously ENABLING advanced Snapshot and FlexClone functionality.  All of this is offered at peak performance via automatic dynamic real-time disk tuning (courtesy of WAFL's Write-Anywhere enabling technology), as evidenced by our long history of remarkably consistent benchmark results.

The Proof

Most of the customer success stories on NetApp's website feature overall efficiency as the reason for switching from competitive systems over to NetApp, but one of the most topical stories is currently featured on Wikibon.  Note the following familiar theme quoted by the customer:

(increased) 20% storage utilization to 70%

The Big Picture

If nothing else, EMC has scored a moral victory by temporarily narrowing the scope of the discussion to what a single array can do using only conventional technology of simple LUN's and generic snaps.

But ... what about thin cloning (aka NetApp FlexClones) technology deployed on a massive scale?  What if you could safely combine it with space-efficient RAID and Pervasive Deduplication with no impact on performance?  Imagine if you could deploy dozens of petabytes of virtual capacity with the same performance, availability and eco profile of a single petabyte?

Well imagine no more. This is exactly what the following enterprise NetApp customers pictured below are doing TODAY.  There are some incredible stories behind and each and everyone of these deployments which I or your local NetApp sales team would be happy to discuss in more detail.  All you have to do is ask!

Comments

[EnterpriseStorageGuy]

Do you have more information on the multi-PB NetApp deployments @ Oracle or SAP?

[the storage anarchist]

I doubt that you have the right-to-use permission from all of those companies whose logos you're flying so proudly - most of their legal departments prohibit any use of their name or logo as an implied or implicit product endorsement.

I know, because almost every one of them also use EMC gear.

[Val Bercovici]

Hi Anarchist,

I know it's a scary notion for EMC that NetApp has such a big footprint in name brand accounts where you've enjoyed a monopoly in the past, but it's part of the new reality.

FWIW - All of those company names and logos also appear in customer case studies on our external web. Often where EMC infrastructure was replaced! :)

However, if our legal folks have any problem with this image I will of course remove it from this post.

[Val Bercovici]

Hi EnterpriseStorageGuy,

Great question! I'm compiling some exciting info about this Month's Oracle OpenWorld which I will share later today. Stay tuned. :)

[Val Bercovici]

Hi EnterpriseStorageGuy,

(thanks for your patience, the late breaking news is at the end)

A partial list of information about our substantial deployments at SAP and Oracle are available here:
- Oracle Implements NetApp Storage for Its Global IT Infrastructure
- Oracle GIT To Save $2 Million With Oracle Enterprise Manager Grid Control and NetApp
- Exchange and Archive Consolidation at SAP Hosting (BTW more details on SAP’s cloud deployments coming soon!)

However what I personally find more interesting and gratifying is the true level of partnership we have the privilege of enjoying with both of those IT staples.

For example, Oracle’s and SAP’s architects and product developers actually co-designed our innovative FlexClone (thin cloning aka “writable snapshots”) functionality. Their deep involvement with our own engineers throughout the 2 year development, testing and qualification cycles helped make FlexClones (and FlexVols thin provisioning as a result) the massive success they are today with tens of thousands of deployments.
http://blogs.netapp.com/dave/2007/12/test-and-develo.html

Oracle themselves now have about 12PB raw in just one location, but they consider it over 60TB usable due to FlexClone innovation!
http://www.networkworld.com/supp/2005/ndc3/052305ndcbcaward.html

True business and technical partnerships are also why Oracle decided to directly integrate NFS into their flagship 11g database
http://blogs.netapp.com/dave/2007/08/oracle-optimize.html

We also accomplished what I consider the impossible – getting SAP to play nice with Oracle and vice-versa. See this upcoming session @ Oracle OpenWorld as proof :)

http://www.oracle.com/openworld/2008/agenda.html#monday (note opening Keynotes!)

This is a real-world cloud computing implementation by a major NetApp customer T-Systems, running world's largest SAP installation with an ~8 TB Oracle 10g RAC database on the T-Systems Dynamic Services platform with NetApp

The system has been in production since end of January 2008 and generating around 500 million payment transactions per year for around 40 million customer accounts.

The session will presented at Oracle World 2008 in San Francisco, Monday 9/22, 13:00-14:00, Room 102.

[EnterpriseStorageGuy]

Thanks Val.

I'm curious about one thing though. Why is this impressive info buried here in the comments when you could be featuring it in a blog?

[StorageJack]

Seems to me those "logo'ed" customers above are too ashamed to permit EMC to display their names, whereas they are proud enough of their NetApp infrastructure to permit it?