Utility computing vs the electricity grid

Posted on February 10, 2008
Filed Under Computing & Software Systems, Networks | 3 Comments

I was watching Steve Paikin’s Agenda the other day and they had a panel discussing ‘the cloud’ which seemed to the buzzword for what the internet presumably becomes once utility computing becomes the norm. They had several panelists on the show – Jesse Hirsh, Nicholas Carr, Shane Shick and Mathew Ingram.

Adoption of ‘The Cloud’ was often compared to a similar transition which powered the industrial age: Just as businesses eventually tossed their in-house generators and adopted use of the electricity grid, so will modern businesses (and households) toss their data centers (and bulky PCs) for a brand new computing utility.

This metaphor is used so frequently that I started to get curious about how the current (ha-ha) electricity grid actually works, and whether some of it’s attributes will be shared by the new paradigm of utility computing.

Regulation: Hmm… the electricity grid is was mostly regulated, but regulation on the internet doesn’t work very well. Most of the big players (to quote examples given by the panellists: Google, yahoo, microsoft, amazon etc) are after our data, fawningly proposing to store it, nurse it, back it up and make it available whenever we need it. Yet it’s never very clear how they achieve this, or where the data is actually physically located. Also, utilities aren’t security-sensitive commodities (other than where usage and billing records are concerned). But with online applications that upload and store our data, clearly there’s an issue.
The power plant: The grid is just the medium for bringing juice from the power plant to the the end-user. With utility computing, the grid is therefore the internet itself, with the ‘power plant’ being equivalent to a massive data center. HOWEVER: since these data centers are themselves massive power sinks, they are going to have start co-locating (and dare I say, co-evolving) with power plants.

Blackouts: An outage in the new-fangled ‘cloud’ would leave you feeling not just frustrated, but rather silly. After all you would suddenly have lost access to your own data. Also, unlike blackouts of the electromagnetic kind, people who lose data and services in the cloud may be spread out geographically – maybe even in different parts of the world (unlikely maybe, since telcos will jump onto this bandwagon, and they have a ‘localising’ effect – but still…). Note that utilities companies have been trained (forced?) to pander to the consumer with regard to blackouts – they offer advice and friendly reminders about candles and flashlights. I have difficulty imagining the computing ‘cloud’ even pretending to be accountable.

While SLAs (service-level agreements) might be in place for businesses when they enter into contracts with their data centers, I can imagine several nifty EULAs and Terms of Service getting between Joe public and their service provider of choice when things go wrong. Words like ’service is provided as-is’ and ‘you agree to indemnify –’ etc.

Metering: Well, there’d only be one way to make money as a service provider / data center in the looming scenario: everyday folks will have to eyeball ads, as usual, in order to get services for free, and businesses will have to pay monthly subscriptions (only fair, since they’re being saved a tremendous amount of hassle – or at least that’s how the sales pitch will go). But this is going to lead to a terrible stratification of the internet. Once the big players get used to different quality-of-service offerings, it won’t take long to connect the dots and just have a whole different net available to different demographics. :-(

Marketplace: there’s a wholesale electricity market replete with futures, options and God knows what other financial gadgetry, in spite of the fact that electricity can’t be stored: the supply has to vary to meet the demand, in real-time, leading to some very oscillatory prices at peak-times. I wouldn’t be at all surprised if a few large data centers – large after swallowing up all the smaller players, no doubt – started hot-swapping resources to keep things on an even keel: a bit of extra bandwidth from these folks, extra virtual servers from those folks, and so on. If data centers think they’ll be able to keep capacity in one place, they could learn a thing or two from the utilities folks.

In short, the analogy with existing utilities is probably a pretty good one, and it only falls apart where privacy and security are concerned. I love the idea of having a thin client as an interface to a vast pool of resources, but I’m a bit cagey about handing over my data.

What am I talking about! I have have a hotmail and a gmail account…

Tenuously Linked (blame my tagging):

• virulent clouds: crimeware-as-a-service
• An hour in the sky: Google Wave, Sound Cloud.
• utility socialising : the future of social media?
• Sites in the cloud
• Databases in the cloud

Comments