C-Special 4: Stacks For Clouds — Will Modular Kitchens Work?

INDIA: MINUTES are ticking by. The grand Super-Chef title is just a few spoons away. The final two contenders hide the tempest of nervous excitement behind busy hands ferociously whipping up stuff in their bowls and pans. And yet, the supporters cheering aloud from the audience area have polarized towards the 30 year old geek who is vying a 55 year old housewife for the golden apron.

For them, what makes the contest tilt towards a blow-away win for the young nerd is a simple three letter word- Can.

A can of garlic and ginger paste that she used while her rival splurged 15 expensive minutes to peel and grind things in a mortar.

A mumpsimus as she might sound, this is how the housewife explained her rationale softly but firmly to the mike-wielding anchor who is jogging between the two workstations excitedly, “I can not trust the proportions and combinations that preserved instant kitchen stuff packs. Then I have my own special spices to go in at the right time. Plus, the caned stuff won’t lift the flavours if my serving size goes up or down.”

“But Madam, don’t you think time is of essence here. Why to waste time in doing something someone else can hand you over.” The anchor shot back.

“Oh, come on, you people only talk about bespoke cars, bikes and garments. Isn’t this the place where ‘your own way’ matters most- in your plate, in your kitchen?” she retorted.

“And that’s why it’s now a battle between a bottle and a pestle,” winked the anchor into the camera.

Ok, admitted that the analogy of cake-mixes, canned kitchen pastes and bottled curries is a bit off the nail here. But the idea and the acceptance (or resistance) of pre-packaged stacks is not that out-of-tune. Why would or why would not an IT manager/engineer allow a stack inside his/her garage follows quintessentially the same flavour of questions that stops a seasoned cook from trading her/his pestle for scissors.

And clearly the big question today is — Would the cans work?

So, what is a ‘Stack’ after all?

VCE’s Vblock, NetApp’s  IVA & SMT (Secure Multi-Tenancy), IBM’s Cloud, HP Blade Matrix in industry vocabulary. But to keep it simple, a ‘stack’ is a canned architecture solution for anyone wishing to deploy their IT and business engines on their own Cloud.

Some describe it as a pre-validated standard implementation, some say it is a pre-integrated infrastructure for virtualization at scale, and others explain it as an out of box ready-to-deploy offering for the Cloud.

The point to note is that stack architectures are gaining a lot of interest, for now in the war-rooms of major Cloud and Virtualisation vendors.

The idea is to heal customer pain points through the concept of "stacks", and that means packing together Servers, Network, Virtualization, and Storage into a single solution. In turn, the stack is supposed to trim off all the complexity and unnecessary engineering time that IT plumbing for Cloud takes.

A simple way of tying up all the requisite technology to work in sync as a datacenter platform, for a Cloud and serving it up in a ‘ready-to-eat’ plate for the customer. Apart from components like Storage, Networking and Servers the ‘stacks’ also include management and orchestration parts like integrated APIs and tools.

Either the customers can build their own stacks, thereby the engineering burden or they can purchase a stack from a vendor.  Here too some vendor stacks could resemble a reference architecture, which basically translates into existing components certified to work together in a specific configuration. Or vendor stacks might possess characteristics of a more cohesive product.

{#PageBreak#}

Off-the-shelf stuff! But why?

A glance at VCE Coalition's goal of Vblock explains it adequately.

The objective is to provide pre-integrated and validated solutions with a guaranteed service level.

Because it frees the customer from the configuration burden. The process to deploy applications becomes far more rapid.

Add to it the benefit called standardization which quickly manifests as any engineer’s dream word — predictability. From power, space, cooling, fault to isolation, it can help a lot in growth planning if things don’t pop out as a bolt from the blue.

Chuck Hollis lauds many advantages in his blog. The CTO of EMC Corporation underlines that there’s a lot riding as bright promise of stacked solutions like VBlock.

And it’s not just about the technology. You cannot run a virtualized Next Generation Data Center (NGDC) using the same processes that were used in a physical data center — and few customers would have the time or resources to develop those practices.

It turns out that the promise of ‘virtualisation’ is already drilled deep and needs no more selling with so many customers raring to go for it to reduce the cost of computing.  But ‘getting-ready’ type of obstacles stand stubborn and unrelenting like a Father-of-the-Bride. This is where chaperones like ‘stacks’ come in.

It might also rub some glitter on the wallet side.

With the advent of ‘stack’, there’s a new procurement model because the unit of acquisition/warranty is integrated infrastructure, not piece parts.

There’s also the single-vendor support model as opposed to multi-vendor joint escalation.

{#PageBreak#}

The other side — Stacks have sharp edges

But then why criticisms stack up higher for these ‘boxes’ than how the benefits pile up?

A lot of brickbats are flying in hordes towards these products.

First, the obvious argument.

Why to go for a pre-configured ‘stack’, when a customer can configure the pile on his own? Just pick up your basket and mix and match Server, Network, Storage and Management tool server. Tada! There’s your own homebrewn stack out of any set of parts (e.g. VMware, HP servers, Cisco switches, EMC or NetApp storage, whatever).

Flexibility to have as many pieces you want. For the cherries on top, there’s  interoperability too.

Between a buffet and Ala Carte, the balances can yo-yo quite well.

But then -Life is about choices, as Manoj Chugh President of EMC India & SAARC and the Director of Global Accounts, Asia Pacific & Japan (APJ), for EMC Corporation agrees.

“A CIO has choice. A pre-engineered solution that is tested for specific workloads is here. It has been tested on a reasonable number of machines with some scale and network requisites as applicable. But there might be users who would want to build their own solution. It can be Buffet or Ala Carte. You may want to take and crumble and whip up in your own way with best-of-breed components. We can just offer choices.” Explains Chugh.

Sudev Muthya, President, IT Business CMS Infosystems feels that a set meal should take care of what you want. “It is always better to consume it as services rather than canned stuff, whether I let one vendor provide it or more than one player takes care of it.”

When it comes to applications as a service, he argues that they should be able to integrate it properly.

“Every company has got something different from another company, so some features expected would be unique and not vanilla ones. It would be as per every one’s own requirements. Not all products would serve everything. That’s why a SI is still around.” He points out.

Interestingly, at least in the physical world, people have reconciled their expectations, but perception-setting for Cloud is too awry as Muthya rightly states.

Talking of perceptions, many criticisms hurled at some of these canned products could be real or perceived, but they are still around.

The products that have come out are still trying to eke it out stirring the broth to customer’s expectations.

For example- VBlock has been doled out charges of poor integration, inadequate flexibility and not-at-par interoperability.

Now that Vblock has been criticized to be less open and less flexible than other stacks. It is even suggested by some that in most cases it would likely be quicker to develop a reference architecture than develop an entire product.

In general, ‘stacks’ face challenges on many areas viz. : simplicity of procurement, scope and terms of warranty, single versus multiple points of support , simplification of the stack’s lifecycle (deploying, maintaining, refreshing), elasticity of the stack, an  a customer’s lack of choice about buying more than one’s needs. 

Oracle’s Exalogic, for instance, has received some criticism too. Like that of resembling a mainframe and lacking elasticity, in a  relative sense to other stacks.

‘Stacks’ are primarily differentiated by the class of services that they provide. The system can go out-of-balance if there’s too much of one resource to be useful to the system.

“You can't just purchase an SMT solution, you have to design it.  This is both SMT's greatest strength and weakness.  It is very flexible but it requires a team with the knowledge to put all the pieces together.” Writes Aaron Delp in his blog, where he paints a quick comparison between NetApp's SMT and the VCE Coalition's (VMware, Cisco & EMC) Vblock solutions.

Advocates of SMT sing about it being a reference architecture and framework that is very flexible and can be implemented into an environment in pieces using common technology, he explains.   But then, it turns out that SMT is designed with a specific solution in mind. 

Those who espouse VBlock on the other hand chant other features, like it’s still a ‘built-in’ solution rather than ‘bolted-on’ rivals.

“Vblock is a product, not an architecture.  It is purchased and delivered in a ‘box’ that is configured and ready to go.  There is a misconception that Vblock is a black box of VMs' that you have little or no control over.”

It ain’t over till the fat lady sings

Now, that will take some time and lot of real-use verdicts as they come out.

But till then, it helps to turn to someone whose sharp eyes watch the stage closely, objectively and astutely enough.

Dennis Drogseth, Vice President, Enterprise Management Associates, Inc thinks it’s too soon to make conclusions about Vblock’s longer term effectiveness.  He would still describe it as an innovative approach to dynamic infrastructure and expect that there will be many related follow ons. 

“However, I also think that some of the failure is the notion that the infrastructure defines the ‘end of the journey’ — It is really just an integrated capability that becomes an enabler in itself and needs to integrate with and become supported by the larger tenants of service management or BSM — to align cloud services and infrastructures to meaningful outcomes and assure and govern performance and change.”

Aaron Delp says it remarkably in his blog: The one big criticism of Vblock I have heard to date is what do you do with it?  SMT solves a technical problem with a solution; Vblock on the other hand provides preconfigured resources.  That is great if preconfigured resources is the problem you are trying to solve.

{#PageBreak#}

“I would love to see the VCE coalition's message improve here.  Right now the message is ‘preconfigured resources’ but I think it needs to be more than that; we need to use Vblock to provide solutions to customers, not just technology.” He suggests well.

Till then, NetApp is rolling out FlexPod storage, network and virtualisation bundles to counter EMC's Vblock bundles.

And why not, as a consultant highlighted in a media report about FlexPod’s launch, EMC is generating a lot of business through its Vblock bundles.

Where will the stacks go?

As an expert rightly says: ‘Stacks’ could be just a new way to sell solutions at a higher level. But then the catch is that while many architects get caught up in selling and integrating the pieces, the clincher is what (and how soon or well) the CEOs and CIOs want to deliver to their customers.  “The pieces of the puzzle doesn't matter as much as the mapping of the solution to the business.”

Aaron Delp quips: It's no secret that I'm a fan of stack infrastructures.  I firmly believe that by helping customers remove some of the installation and configuration (nerd knobs) and creating a solution that is certified and proven to run your customers' applications, you create a new way to serve your customers in today's evolving industry.

Listening to so many perspectives leaves me confused and curious.

It turns out that in Cloud, ‘stacks’ can cover a new stratosphere if they wake up to and address the issues that customers are talking about.

Yes, it could be an individual customer’s choice of build Vs buy at the end of the day. Unless ‘stacks’ succeed in adding a new factor to the equation — build Vs heat-up Vs bake.

Doesn’t Dennis help with a wiser, top-of-the-woods view here? The “journey to the cloud” is itself a meaningless statement, as he opines.

“Cloud is merely a confluence of enabling technologies that are also a catalyst to some of the organizational and role changes but cloud does not define this.  IT’s emerging role as a broker of services and business model transformer is in the end transcendent to cloud.”

With so much discussed and debated about canned products, the book still is far from closed.

Because even today, a Harley owner takes greater pride in carrying his/her toolbox along rather than a mechanic’s phone number.

Back there, the dishes have just arrived on the table for Judges.

And even though the crowd feels that the housewife has lost thanks to her quirks, everyone is waiting for the results with bated breath.

And, the award goes to……..