« About Dan Oprescu |
Main
| Dealing with the limits of science »
October 19, 2009
Risks in the Cloud
Cloud computing has made the front page of The Economist last week, after years of steady progress. It started out as a revolutionary way of performing everyday computing tasks, and now the big firms are battling for supremacy in this area – and indeed, for the future of computing, many billions of revenue for years to come, and for some, perhaps their very survival. Everything will be thrown into battle, from the most modern viral marketing to classic legal action; the consumer will just have to stand back and wait for the winner to emerge…
On the face of it, who could argue with the cost advantages and the convenience of cloud computing? Desktop computing – a manufactured commodity product – is being replaced by services delivered by a utility. But let’s examine the risks:
1. the services require something that the customer owns: data. How is cloud computing protecting data ownership and the value to their owner? Sensitive information will still have to stay ‘in-house’ throughout the entire workflow. A version of this risk is particularly worrying – the State can invoke the national interest, and information which in the past had to be gathered is now all in one place, ready for analysis and manipulation.
2. one of the strengths of the Internet was its built-in redundancy. But in an ideal world (for the giant providers of cloud computing), a small number of companies will share the market between them; should any of them fail for any reason, large numbers of customers will be left without even the basic computing capabilities. The comparison with utilities is particularly apt here, remembering the massive blackouts which happened in recent history on the East Coast of the US and Canada.
3. recent cyber attacks have exposed vulnerabilities unknown before; as computing power is concentrated in the massive data centres that the big players are building, it will become so much easier to knock out large parts of the economy.
The list can go on, but let’s see if we can quantify the risk, and whether the savings are worth the price of risk. In one scenario, what is the financial impact of a total computing blackout of a bank, lasting 8 hours? It can be modelled with one of the accepted stochastic distributions for loss severity, like the lognormal distribution, and for the sake of this argument let’s say that the expected loss would be 1 day profits . The probability of impact is low, to be sure, hard to estimate from historic data; what if we take it as a one in 10 year event? The resulting loss would be 1 day profits in 10 years of operation. Is it worth all the savings brought by The Cloud?
Posted by jaguar36 at October 19, 2009 09:47 PM
Post a comment
