Overtake this! Symmetrix V-Max? Frames are so 90’s

Wither the DMX moniker? With all his legal troubles, I guess the name association was getting a bit troublesome :)

Without too much digging tonight, I stumbled upon an over-view of EMC’s over-hyped “Over-take the Future” product announcement set for tomorrow. Seems like FT didn’t quite plug all their leaks of the EMC press embargo.

3Par’s Deep Throat and Rickroller extraordinaire Marc Farley has a point that EMC seems to be plagiarizing the InServ spec sheet from a product perspective. Seems like Compellent will soon be releasing a press release quoting their CEO with the same opinion. But that box-oriented viewpoint is far too narrow.

Does the storage world need another Hummer?

The Data Center of the Future is about far more than a new frame array or a new firmware feature. Non-invasive enterprise infrastructure discovery, rapid provisioning, scalability, automation, policy-based management within a standard ITIL / ITSM framework is what’s needed to get there.

And yet the advent of Cloud Computing dwarfs individual data center concerns by shifting the landscape to virtualizing all of IT as a granular personalized service.

At NetApp, this evolution of thought has been underway for some time now. If you’re curious what we can do for you today (or if you’re a competitor, just how far NetApp is ahead of you) feel free to browse our thought-leading:

Dynamic Data Center blog

Is It Compatible?

In this economy, EMC, please don’t try to sell me more new stuff when I can barely afford to keep and manage my current stuff.  If you can’t let me improve, protect and extend my investment in the storage system I already have (which could include the one you previously sold me) then we have nothing more to talk about.

Is it Unified?

Integrated native file, object and block services with a single consistent management interface is what I mean. Despite the empty Celerra marketing claims, EMC doesn’t ship this kind of solution.

“Purpose Built” is the antithesis of “Unified”. And those who only claim Unified Storage for one isolated end of their product spectrum are completely missing the point.

Unified Storage is not just about protocols. It’s also about unified discovery, provisioning, management, protection, archiving and DR. A Chinese menu of purpose-built solutions won’t get you there.

Is it even Guaranteed?

Innovation is not restricted to technology. Why not innovate in the introduction and delivery of the overall solution? Why not prove the value and efficiency of a proposed solution in the most transparent and material manner possible?

The media is noticing, while Influential analysts are not only universally recommending all virtualization users demand these programs - they also skillfully defend them when attacked by biased heretics.

Now that VMware itself has seen the light, is there really even a debate any more?

Why won’t EMC back all this product hype by putting their money where their marketing mouth is? Prospective Symmetrix V-Max customers should demand more than business as usual and the same old promises. They should:

Demand efficiency guarantees!

Oh, and for those captivated by 1TB of global array cache

Be prepared to blow your mind.  Without giving too much away, our current mid-range and high-end systems are already capable of hosting >1TB of cache per node.  This functionality will be enabled by our 2nd generation Performance Accelerator Module (aka PAM) due later this year.

Let’s Play Jeopardy.  The Answer is “40TB”

What is the effective maximum usable cache available in a high-end Active/Active NetApp FAS or V-Series array Alex?

That is correct.  This question and answer brought to you by NetApp’s exclusive latency-optimized primary dedupe functionality. Which also happens to be market leading. Remember folks, NetApp’s PAM-enhanced implementation of primary dedupe performs faster than our conventional industry-leading non-deduped (i.e. “duped”) primary storage.

Today more than ever - Less (Tiers) is More

Therefore, when you can cache your entire (hot data) working set for performance, why bother storing cold primary data across multiple (FC & SAS) disk tiers? Why not leverage the undeniable CapEx & OpEx advantages of low-cost dense SATA storage for all your primary storage?

Only NetApp proves you can.

That type of storage configuration gives you a single tier to “manage”. From an HSM perspective, a cache-centric architecture is like “H & S without the M”.  After all, modern memory paging and disk caching algorithms are the most proven and robust forms of automated hierarchical I/O management commercially available today.

Move on

It’s time to stop improving the “storage hummer” and move on to addressing this century’s business problems using modern guaranteed, Unified solutions that break the frame array mould!

Comments

[Larry]

Hopefully I can sneak in the first comment before Chuck Hollis or Barry Burke.

If so, I'd like Chuck to answer why EMC isn't confident enough to guarantee the efficiency of their products when VMware is? And Chuck, please don't try to patronize me about how much more complicated storage virtualization is than server virtualization. NetApp has not only figured out how to guarantee the efficiency of their own storage - they can now do it for EMC storage as well!

Barry (or Anarchist or whatever you prefer to be called) - kudos on launching yet another Symm. Can you answer why I have to forklift my Symm's yet again to use this latest round of sexy new features? How about why I still can't connect it to the other arrays I already own, or when it will finally support NAS naively within the same management paradigm? After all I hear from Chad that NFS is the way to go for VMware!

[Jeff]

Val, you keep mentioning provisioning, policy-management management, replication etc. Where does orchestration factor in? Isn't that the ultimate requirement for successful cloud deployments?

[Roger]

Seems like an impressive feat of HW engineering by the Symmetrix Engineering team. OTOH - This product was clearly conceived during the heydays of the market boom and is therefore completely inappropriate for these lean times.

I see no innovations around efficiency. Nothing around low-cost entry level systems, nothing around low-cost granular growth increments and nothing around improving the efficiency of the rest of my infrastrcuture. (iSCSI, FCoE, NFS, IB, iWARP)

While VMware provisioning promises to be simpler, what about my HYPER-V deployment? What about KVM or Xen?

Can you imagine Amazon standardizing EC2 or S3 around this kind of architecture? Neither can I - and that will be the death-knell for this "bigger is better" type of product architectures.

[Marc Farley]

That is one beautiful striped zummer (cross between a zebra and a hummer) you got there! Faster than a Zorse or a Zonkey and holds more people and only slightly less of a freak show attraction.

[ObjectCurious]

Unified *object* interface? I understand EMC has no such beast, but can you tell me more about NetApp's as you hint above?

[Vincent]

Hey EMC, it's 2004 and they want their roadmap back.

Seriously, didn't 3Par ship this about 5 years ago?

And 40TB of CACHE on a filer? Is that a typo or are you sh**ing me?

[GeorgeV]

The hardware engineering (RapidIO, Intel CPU's) is indeed impressive. But are we seriously to believe that the Data Center of the Future will be based on 4Gbps Fibre Channel and single GigE?

Talk about a performance, operational and strategic bottleneck! Where is 10GbE and the much-vaunted FCoE? While we're at it, NFS over 10GbE and QDR IB would not only be nice, but arguably required to realize the full potential of all that controller horsepower.

Lower-cost SAS drives for Tier1 seem to be MIA as well. RAID levels continue to remained mired in the 90's. Can't wait to see if the reference architectures ever sway from the recommended RAID10.

And sadly the coolest parts of all (De-dupe, FAST, Thin Provisioning, Federated Management, QoS) seem to be mostly pre-announcements with very little meat.

Overall this Hardware-centric announcement feels rushed. I wonder if it's due to a weak Q1 and/or ongoing pressure from HDS & 3Par - and of course NetApp?

[Paul P]

As a famous EMC blogger recently posted, and I'll paraphrase: This new piece of tin (which cannot utilise any existing systems) is the same price as a bigger Clariion. So I expect that EMC will soon stop shipping the bigger but limited Clarions.

[John F.]

The last time EMC took firmware and put it on Intel, that was the Clariion. They ported Flare to Windows XP embedded. I wonder what EMC ported Enginuity to; Vista embedded? Anyone know?

[Michael]

I'm a satisfied CLARiiON admin, but feel jilted and left out from all these cool VMX-only announcements :(

Why can't EMC focus on a single, consistent storage family where every customer benefits from their $1+B R&D spending?

The $250K ante to get these VMX features will always be too rich for our blood.