We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.

The Connected Web

Phil Wainewright

Most Clouds are Too Clever By Half

Vote 0 Votes

It's been a big week for cloud computing, but I can't help thinking that most players are missing the point. The cloud — what we used to call utility computing, remember? — is fundamentally about the commoditization of IT infrastructure. And the crux of commoditization is that simplicity always wins out over sophistication.

This week we've had Cisco coming out with its Unified Computing System for cloud data centers, we've had IBM supposedly in acquisition talks with Sun, and Sun unveiling its own Open Cloud Platform, while Microsoft revealed a little more of its cloud roadmap. Yet what I keep noticing, unheralded by any headlines, is the creeping de-facto standardization of Amazon's cloud. Analyst James Governor hits the nail on the head in his posting this week on Amazon Web Services: an instance of weakness as strength:

"Amazon isn't the de facto standard cloud services provider because it is complex — it is the leader because the company understands simplicity at a deep level, and minimum progress to declare victory."

James' point is that Amazon has taken the bare, simple, inspired minimum to enable cloud computing — while all its competitors are trying to be too clever by half. "The beauty of instance simplicity is ... no new skills," he points out, whereas, "The problem with fabric complexity is the promises being made." In Amazon's own words, it now supports IBM, Adobe, Oracle stacks, as "no-fuss, no-muss application development and deployment."

Or as Nick Carr put it, under the title Another little IBM deal that recalled IBM's coronation of Microsoft when it launched the PC back in 1981: "corporations and software developers will be able to run IBM's commercial software in Amazon's cloud ... Could the accidental kingmaker have struck again?"

Increasingly, I'm seeing evidence of a phenomenon Ismael Ghalimi first predicted more than two years ago: "Imagine for a moment ... deploying clones of Amazon's most popular Web services, including EC2, S3, SQS ... onto your own servers. Imagine that more elaborate versions would be developed on top of some kind of grid operating system ... Finally, imagine ... client-side software ... to use Amazon's Web services in combination with any other clone in order to get failover capabilities, without having to learn any new API."

Underneath this commoditization of the cloud layer, of course something drastic is happening to the underlying infrastructure stack. As Saugatuck Technology notes in a research alert (PDF, registration required) published today:

"... economic conditions and the growth of Cloud Computing are pushing user demand and interest more rapidly from standalone solutions to combined, integrative, cloud-based, or cloud-centered offerings ... The server landscape changes on a massive scale as a result."

This heralds massive consolidation of the server, network, storage and management stack, as heralded by Cisco's announcements this week. John Willis encapsulated it neatly with his observation that, "It's one thing to automate server elasticity; however, it's really big game if you can automate the configuration and provisioning of servers, network, and storage devices all in one swoop." But nevertheless, I can't help feeling that Cisco's all-too-elegant integrated implementation is yet another too-clever-by-half attempt to do something that commodity players will do much more simply, cost-effectively and (crucially) non-proprietarily.

In conclusion, I suspect the key point that most of the players seem to be missing about cloud is the reductionism that a true commodity infrastructure platform requires. It's only when you've stripped out all the non-essential parts and reduced it down to the bare necessities that you reach the ultimate cloud platform. But of course most of the players can't afford to do that — quite literally — because it strips out all the margin that generates their livelihood.

Cisco talks about the cost reductions (and petite carbon footprint) of its new data center platform. But in reality it's barely scratching the surface of what can be achieved when, following down the path it has now opened up, the physical underpinnings of IT infrastructure get fundamentally remade for cloud computing. The outcome of the chain reaction that has been set in train this week is going to drastically rewrite the economics of computing.


"And the crux of commoditization is that simplicity always wins out over sophistication."

Really? In what technology commodity? Certainly not Operating Systems. Or Office Suites. Or Databases (Oracle still on top, still very complex). Or IT Service Management (Tivoli and OpenView still on top, still very complex). At least, when considering "in Enterprises", and in terms of software revenue.

Cloud and virtualization, if anything, are drastically increasing the natural complexity of managing IT, because they enable so much more freedom to be sloppy. Just like PCs were more complex to manage than mainframes, VM and image sprawl have already begun to be the next set of problems to deal with. This won't stop their adoption, but certainly there was a cost to this associated complexity (and much money spent at solutions to reign in the complexity).

The large IT vendors are understandably seeing this as a "Smorgasbord" opportunity, where they can package bundles of 10+ products to solve people's cloud problems. Naturally that's introducing a lot of artificial complexity into the problem, and startup cloud vendors will have a fun time playing catch up but eventually offering a more "essential" solution that is end-to-end.

But here's the catch: Simplicity is not just about reductionism, it's about good design, layered abstraction & modularity, and an end-to-end experience. And it is usually is very, very hard to pull off.

On demand infrastructure like EC2 simplifies a chunk of IT, and is an inspiration to start from -- but is a) quite proprietary, and b) far from the majority of the typical enterprise configuration management problem. More sets of APIs aren't going to solve that problem, and neither will outsourcing to a set of PaaS vendors (we still have a data integration nightmare to deal with).

In other words, there's no free lunch. Some problems, such as IT and business service management, are naturally complex - and even when you reduce the problem into its essentials, there's so much variability and combinatorial possibilities, that the solution is inherently going to be complex.

I expect that there's an opportunity to solve "cloud problems" simply and end-to-end, but I would say it's far from a fait accompli, and the way it will likely happen would be by building some rather sophisticated software that looks & feels simple. Which, as with EC2, or Google, or SalesForce, will have major proprietary elements to them.

if you want to explore Amazon S3 online storage and configure CloudFront CDN I suggest that you try CloudBerry S3 Explorer Freeware http://cloudberrylab.com/

Stu, thanks for leaving such a detailed comment. Unfortunately I disagree with large chunks.

In particular, I found your examples of sophistication (Oracle, Tivoli ...) instructive. Here are platforms that replaced previous generations of technology precisely because they were simpler and far, far cheaper than what had gone before. Of course, 10 and 20 years on they've become bloated, overpriced and due for replacement, but look back to their beginnings and you'll see parallels to what we can now expect in the field of cloud computing.

Client-server, for example, looked like a management nightmare to start with and was a hugely complex proposition for any of the then-established management vendors to deal with. Then along came Tivoli, which IBM ended up buying for $1 billion because the startup was able to solve a problem that IBM with all its expertise and expensive software couldn't get on top of.

In the end, I do agree that the solutions that get established will be "rather sophisticated software that looks & feels simple." However my point is that they'll also be hugely cheaper and less horrendously complex than what we're currently seeing proposed from the likes of Cisco, Sun, IBM and others.

Hi Phil, Stu...

Nice to see our worlds mashing up again...

Rather sophisticated, looks and feels simple--that is EXACTLY THE CASE!

In the interim, Cloud will do more to add complexity to the Enterprise than to reduce it. For green field development in say a startup or SMB, Cloud will help simplify things a lot. For the rest of the world, Cloud = increased complexity.

This is logical and perpetual for IT however. Macroeconomic forces of consolidation and competition in software vendors AND in the broader market demand it. Complexity is here to stay.

My 2 cents,

Phil Wainewright blogs about how businesses are using the Web to get better plugged into today's fast-moving, digital economy.

Phil Wainewright

Phil Wainewright specializes in on-demand services View more

Recently Commented On

Recent Webinars


    Monthly Archives