Tuesday, June 17, 2008

Sustainability: So how do we break the mould?

Sitting here at the BT Centre in London where I am attending the ITU conference on ICTs and Climate Change, see http://www.itu.int/themes/climate/. The conference is focused on the challenge of climate change and the necessary steps to be taken to achieve a balance between man made emissions and a level playing field for developed and developing worlds looking forwards. Most notable during the day has been the sequence of contributions from the hosts as they have presented their results to date. A very creditable and real reduction in carbon footprint. See the presentations, via video and powerpoint, available from the website.

However, that's not the purpose of this post. I also attended the SIIA conference in Amsterdam last week (no jibes about carbon footprint this week please!). I was most impressed by the rapid evolution of SaaS and the very practical steps that people are taking to offer new services and capabilities on-line for consumption via the simple browser. Here was a genuine low-carbon conference dominated by the plans and capacities of an industry structuring itself around Infrastructure as a Service, Software as a Service, Information as a Service. The global champion in the room was Amazon Web Services, again through the person of Werner Vogels, see my previous post. But the buzz and excitement amongst attendees was palpable around the delivery of business services through these mediums. There was even a session focused on Security which was far less frightening than removing an orange folder to take your seat on a train at Waterloo Station! General consensus, this is real - and its changing the game!

So, to my point, we seem to have a created by default a product development cycle and roadmap in the computer industry which assumes: 1) that the next box will be more powerful, have more capacity and probably expend more carbon resources as a result. 2) we will fill this new box with a larger amount of software than we could ever imagine using, and thus negate the benefit of increased power and capacity and 3) will be persuaded to exchange this barely utilised resource in 3 years time because of a step function increase in processing power/storage capacity and an impressively colourful lid.

When you consider that data presented at the conference indicated that desktop pc's consume about the same energy to operate over 3 years as to manufacture initially and then the disposal will consume more, irrespective of potential/process for recycling, the question "WHY?" pops into my head. When you combine this question with the apparent potential of XaaS and perhaps a simple wireless connected browser. I begin to wonder why we are continuing on this treadmill of a roadmap and all of the inerent costs implied.

In one panel session at the ITU conference, we heard from an EC officer about increasing regulation and a trade association representative about allowing market forces to take their effect. At present, we as users, are running fast to continue to buy the newest, more powerful and therefore likely most expensive option to solving their business problem. If you believe in market forces, then I am afraid it is those of us who use systems who will need to drive this change forwards. Who's with me?

Labels: , , , ,

Tuesday, June 03, 2008

Barcelona, Grids and the Cloud

This week is OGF 23 in Barcelona, a global grid community gathering in support of the agenda of grid standards and adoption in the scientific and business communities. The event is also shared with BEinGrid, an European initiative to demonstrate business relevance of grid, and the launch of OGF-EU, an European FP7 project aimed at championing the standards cause and European contributions to the development of grid. I'm pleased to say that my employer is supporting OGF-EU as well as my continued stewardship of the Grid Computing Now! Knowledge Transfer Network.

So what's the buzz, we're two days into the event and a pattern has emerged, right? Well, yes, it has. Simultaenously, we are told that Grid is not being adopted as expected and that Cloud is the answer. We also had an interesting debate following Werner Vogels excellent keynote speech, he's Amazon's CTO, about whether the "cloud" is in the grid or the grid is in the "cloud"!

There are grains of truth in all of these propositions:-

Firstly, grid is not likely to be widespread in its implementation as a technology. Its proposition is that you are able to distribute your applications to operate in parallel in an heterogeneous compute environment, within and possibly, without your organisation boundaries. This of course requires some form of restructuring of applications previously written with specific hardware configurations in mind. Not a likely course of action for most understaffed (in capability and numbers) IT departments. Without doing this the application cannot take advantage of a grid infrastructure. In addition, configuring grid infrastructures to support transaction processing is in its early days, outside those experts in the major application suite suppliers and leading web service providers, all of whom are building their applications to distributed infrastructure architectures. So, the likely range of adopters is limited.

Secondly, cloud is the next big thing and a sexy concept to discuss. The reality is that the leaders in the cloud service provision industry, which certainly includes but is not limited to Amazon with its Web Services, are simply dipping their toes in the water with large scale commodity infrastructures leveraged from their understanding of the their own e-commerce business models. Remember that the likes of Amazon and e-Bay already operate huge e-commerce infrastructures and literally service millions of suppliers, drawn by the lure of customers! However, looking forwards, the scale of infrastructures being built by these leaders and Google and Microsoft and ... is truly gigantic, often environmentally efficient - in a way that those with data centres in the centre of big cities really can't compete with, and they have the capacity to support many customers before they start to sweat on further investment. The rub is that they have invested in low technology commodity hardware and high technology competitive management practices to deliver resilience and efficient operation - all in the name of keeping the cost of services low - and much lower than the owned data centre!

But you say, doesn't security floor this proposition? Yes it does, but not the way you may think. Its governmental and industrial regulation which is providing obstacles to storage of data on non-dmoestic infrastructures; or an audit trail which forces the trading company to invest in masses of expensive, redundant on-line storage. Perhaps there'll be a white knight government or sector who will look again at these restrictions and the added costs they incur.

Finally, to the Grid in the Cloud or Cloud in the Grid debate? Firstly, who cares? It is a trusim that grid computing technologies are at the heart of all of the new "cloud" infrastructures. Its also true that people can implement grid, which means installing and running middleware in cloud instances to form a virtual grid. So you tell me which came first, the grid or the cloud!

The bottom line of all this? What price the owned data centre of the future. Is it possible that wholesale change will impact the enterprise data centre as CFOs struggle to reconcile a cost m odel which is 10x that of the leading cloud service providers? Who else will be there? Well, the telco's are looking at network service provision with an eager eye. The Big tin vendors are increasingly being pushed down the path of adopting large scale customer infrastructures. Where might they make their profit? In the Cloud!

Labels: , , ,