Cloud Cuckooland economics ?

From ZDNet comes news of Forrester's thoughts on why a company should use the Cloud rather than its own servers (I was going to comment on ZDNet, but their 2 page commenting signup process made me decide to blog it instead). Anyway, below is the economic argument:



Although this is all very true, unfortunately this is not the full picture of the real economics of Cloud computing. Here's why:

The sums you spend on The Cloud bit of the product or service are usually quite a small % of total value, unless the service is very processor / bandwidth intensive for its price (think YouTube). This means that the savings here are trivial compared to the risks you face if the Cloud:

1. Falls over and cannot serve your customers - in which case you stop paying the Cloud its pennies (hopefully you even have a few penalty clauses) but you lose Pounds.

2. In any way causes your confidential data to be lost, stolen or otherwise mishandled

3. Believes you have in some way abrogated its T&C and suspends your service.

To mitigate these you will need to ensure the Cloud contract also provides you with:

- A Service Level Agreement (SLA) that guarantees a level of service (In the Old Networked Economy the Gold Standard was 99.999% uptime) plus a rate of response and penalty clauses that, if valid, motivate the provider to get you back up darn fast.

- An End User Licence Agreement (EULA) that makes it VERY clear that the data is yours, irrevocably, and cannot be used / sold / mishandled / stopped etc without VERY major damages to be paid. It is interesting to look at the EULA's of some current low end Cloud providers and seeing just who owns your data - a very recommended exercise for any would-be buyer.

- A Plan B, so if The Cloud in any way fails, you can switch over in a heartbeat to a working service. If this cannot be done, it requires you to pretty much run your own services, which means you may as well do some of it yourself - or back it off to a 2nd provider- in which case see if they can guarantee uptimes

These are the Real requirements for a Cloud service to be worth using for companies where the added value is high.

Now there are companies that can use it now, without these conditions being met - typically those with very high reliance on the infrastructure efficiency at this level, and where loss of service differences between value added and cloud costs are not too wide, and the added value of the content is not very high. Startups and processor hungry social media / user generated content plays fit this bill. For others, one really needs to get the sort of SLA and EULA that is common in traditional outsourcing to make it worthwhile.

I hate having to be the curmudgeon, but we are still firmly in the Hype stage of the Cloud (Gartner says so, so there!) and we probably need it to sink through its Slough of Despond before most commercial companies should look at it seriously - by then the cowboys will have gone out of business, the learning curve / Moore's Law effects will have made the services more reliable and economical, and there will be enough history with existing providers to know who is competent.

Until then, buy a spotty youth, a big server and a LAMP manual

Update - I see this this post has attracted the worst karma ratings since I had a go at Twitter in its geek first love phase - clearly Cloud is the new black (as broadstuff's writing and intellectual rigour is always of course above reproach, that cannot be the issue )