Virtualization Technology News and Information
Cloud Computing and How it Could Hurt your 2011

Cloud Computing and How it Could Hurt your 2011

Contributed Article by Guise Bule, CEO of tuCloud Global

For most of 2010, I avoided talking about the cloud as much as I possibly could and felt ever more embarrassed as the concept of the cloud pierced even the consciousness of even my own technically illiterate parents.

I say embarrassed because when we named our company tuCloud back in 2007, it seemed like a good idea at the time, the term had not yet been prostituted by anyone with some kind of service that somehow connected to the interwebs.

By the time I got to 2010 I had begun to have serious thoughts about renaming my company. The word cloud had become so debased at that point I thought all was lost, but then I started to think about how far cloud computing technology had actually come since 2007 in terms of actually being able to live up to the promise of 'the cloud' and I can tell you that in 2011, a number of very interesting cloud architectures will compete for traction and credibility.

I do not want to talk about specific vendors cloud plays here, instead I want to explain what cloud means to me and then go on to explain what I like about where I think the cloud is going.

Cloud of course was always going to be one of those words that permeated other languages without morphing into something more local. In 2011 I was in Spain, Portugal, Germany, Italy, Denmark and the Netherlands, all they wanted to talk to me about was "the cloud". NOT el nube (spanish), le nuage (french), die wolke (german) or even a nuvem (portuguese). 

The Cloud.

When a Portuguese fisherman's nods wisely at you and tells you that the cloud is the future without being able to quite explain why, you know you are in the right business.

In 2011, I have decided to stop trying to irrationally fight the cloud and go with the flow.

I will accept that in 2010 I lost control over the definition of 'the cloud' to the desperate definitions of a thousand others who had sprung up during the course of 2010 to become 'cloud experts' and explain what they thought The Cloud was.

In 2011, I am all about the cloud, but not in the bull**** sense, I am all about the cloud in the actual expecting something from it sense and I look around and I can see that a lot of you are thinking the same.

What do we want from the cloud ?   Well we want to save money, time, be more efficient in the way we do things and we want it to be shiny and new (cloudlike ?).

For our users, the cloud will always be a concept, a brand or a marketing term to be assaulted with, if you are in the cloud business and you look around and all you can see is concepts, brands and marketing then you know you're in trouble.

The cloud is really about the datacenter (DC). Every DC is in some way a service provider and from the service providers point of view, the general goal is to squeeze more efficiency out of data center resources and operations and generally we tend to want to go about this via consolidation.

In service provider terms, consolidation means over-subscription of the network and servers. You try to squeeze as many virtual desktops as you can onto a box using a formula based on their resource consumption, but in order to guarantee service levels during peak load times (and build in effective redundancy), you still need a whole big chunk of infrastructure sitting  idle 'just in case'. Your DC always runs at half mast rather than at full steam.

Most cloud service providers with a large user base are in this situation, where they need to have the resource standing ready.  These cloud providers may well be selling cloud to their own customers and from the customers perspective the cloud is exactly what they are getting, it really does do what it says on the tin.

All the customer knows is that today he can have 125 hosted virtual desktops from his cloud provider, but tomorrow if he gets that big contract he can quickly expand to 500 cloud desktops and then shrink back down to 100 desktops twelve months later when his contract expires.

The customer knows that he will pay for what he eats and nothing more and that he gets the resources he needs when he needs it on tap. For that guy the cloud really is living up to the ideal, but the cloud service provider is screwed and generally not feeling the cloud at all.

The key to cloud is adaptive/predictive behaviour, without pre-reservation of resources. If you are pre-reserving resources that sit idle during off peak load periods, then your cloud is really just 'outsourced datacenter 2.0'.

Generally as cloud providers grow their infrastructures, they will simply SCALE UP by adding more servers to the infrastructure. 

So we are clear, that is not the cloud. The Cloud is not scaling up, it is scaling up and scaling OUT.

The most interesting technologies I see in the desktop virtualization and cloud computing spaces are network distributed ones, I get excited when I hear the words 'symmetric scalability' and 'switch cloud fabric'. 

Look out for those words in 2011, proper cloud words they are.

Right now Moore's law is thrashing us with a vengeance, chip caches aren't scaling with processor frequency, which means that sometime in the next 3 years there will be a performance bubble. Scaling processor clock will not translate into higher performance, 'scaling up' will hit a brick wall in other words.

To get more performance in the here and now in order to feed the cloud, we have to use infrastructure architectures that leverage the network in a new way and in 2011, it will turn out that the network is in fact the computer after all.

As cloud computing really begins to mature, the industry will revisit network distributed computing (again). Network computing is all about switch fabric and as the network is evolving before our eyes into something that looks like a viable bus, distributed switch fabric clouds are becoming very sexy indeed. 

The 10gE move will start in 2011, causing DC cloud and compute architectures to dramatically change because of the massively scalable 100gE switch fabrics that are just coming onto the market.

Imagine a rack of low cost 1U servers in a full rack, with each having a quad 10gE. Say you have 25 servers in one rack, then that means that you have 1tE (1000gE) at top of rack for each rack. Emerging cloud computing architectures will strive towards distributed compute and storage which will make it possible to sustain 1000s of concurrent 10gE connections between virtual machines and virtual storage resources.

Using the network to distribute resources and load gives us a new way to scale, in other words we can scale out rather than just scaling up and that is what the cloud is all about.

In desktop virtualization the most interesting and potentially disruptive technologies I see are those that lend themselves to this principle of scale out rather than up, using network distributed resource allocation. 

Some have argued long and hard that the VDI model is broken and that server based computing is simply not a sensible proposition for managing and hosting cloud desktops in a world that still lives far short of the internet age and there have been moments in 2010 when the interwebs threw enough challenges at my HVD platform to make me believe that the VDI model was in fact fundamentally flawed.  

Now that the cloud madness of 2010 has cleared and I am a little wiser, I realize that in the cloud there is in fact a place for almost everything and everyone, but I do think that if the cloud does properly take off and begin its ascent in 2011, it will be scaling out rather than scaling up and this is doubly so for desktop virtualization architecture

So how can the cloud hurt you in 2011 ?  By leading you into choosing a scale up, rather than a scale out architecture, something that would be quite painful as the volume of users on your cloud grows.

Cloud is just as much about scale out as well as scale up and this point will hit home hard to those of us that actually have to invest in cloud infrastructure.

About Guise Bule
Guise Bule is the CEO of tuCloud Global, a global provider of hosted virtual desktops and desktop virtualization consultancy.  Guise writes virtual desktop reviews and talks regularly on the subject of virtual desktops and considers himself to be a bit of an expert on the subject of hosted virtual desktops and cloud computing, even if he does say so himself.

Published Tuesday, January 25, 2011 5:00 AM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<January 2011>