Raining on the cloud

The cloud will go down in 2010; that's another of Mark Anderson's predictions for this year. In his words, "There will be a Cloud catastrophe in 2010 that limits Cloud growth by raising security issues and restricting enterprise trust. CIOs will see the Cloud as the doorstep for industrial espionage."

What's going to go wrong? "It could be either a large service delivery uptime issue or a big security issues - and CIOs will be reinforced in their preference for building their own data centres as opposed to shipping the crown jewels of the business out to Amazon."

And the result? "That the cloud rather than being perceived as the ultimate destination for all will be a consumer thing."

That's both a bold prediction and a very tame prediction. Any time a technology becomes popular, there are going to be problems and cloud is getting popular enough that enough companies are going to be using it this year for us to get the critical mass for a really big problem. It's inevitable - but will it really put people off? After all, there are have countless significant breaches of privacy and data security disclosed over the last few years and we haven’t seen mass adoption of encryption or endpoint management as a result, because they can be complicated and expensive. What the cloud promises is simplicity and cost savings - although Anderson isn't convinced that's always true. He refers to an "interesting startup" that used to be the poster child for Amazon Web Services, used by Amazon as a case study, and no longer uses AWS because of escalating costs.

"It turned out that as they ramped up, costs got out of control and they realised that at some point it would be too expensive to continue." Combine the cost issue and the catastrophe he's expecting and Anderson thinks the appeal of cloud may prove to be limited. "It's a wonderful thing for start-ups, for prototyping... I expect to see a flood of use from new users, but maybe they get to a point where they see the per node costs [going up] and they go back to their own IT."

It isn't magically cheaper to run a cloud than any other kind of IT; the economies of scale and use of cheap, redundant and efficiently managed hardware are what make the difference but the long-term value of cloud platforms hasn't been proved yet - not least because the expertise at getting the most from a cloud platform isn't yet widespread. But cloud is flavour of the month (and in many cases it does seem to be the right answer to a number of IT problems). Will a cloud catastrophe change that?

There's increasing consumer concern about the privacy of their information in the cloud. An entry on a personal blog by Mozilla Director of Community Development Asa Dotzler started an argument last year when Eric Schmidt pointed out that the Patriot Act means Google might have to disclose information it has about its users if the US government tells it to and Dotzler suggested that Bing has a better privacy policy. Schmidt is right about the Patriot Act, and Google's business model means it has to keep as much data about its users as technologically possible, so it can mine the data of connections and opportunities. That's one of the big dangers with clouds - not just whether the data in them is secure from malicious attacks (which businesses have to think about - again, there's nothing magic about clouds and once they're holding valuable data they'll come under attack) but what the privacy and disclosure policies are and what regulations cover it. UK accountants shouldn't use free Gmail accounts to do their business email because it can take client data out of the UK - which the regulations forbid. If your business uses a cloud service for ecommerce or line of business apps and that cloud lives in the US, you don't want to sell - for example - powerful model planes to certain Middle East countries even if that's legal in your own country, because the US considers that having the ecommerce platform in the US means it's US export regulations that apply and will arrest you when you visit the States.

That's already happened. Customer data stored 'in the cloud' has already been lost too. Magnolia, the 'cloud' bookmark service that turned out to be run on a couple of old Macs, the problems with the Danger service that Microsoft runs for T-Mobile US to back up contacts, pictures and everything else on customer Sidekicks… It's going to take something bigger and more catastrophic to turn us off the cloud.

Or we could prepare for the fact that a catastrophe isn't just possible, it's likely and start checking out the security, regulation and robustness of cloud services before something goes badly wrong.

The best way to deal with many of the dangers of the cloud is to treat it as part of the solution rather than the whole answer. We need a better name for this than Microsoft's software plus services, but the idea is sound. Take advantage of local storage, processing and security - on the client or in the data center - but sync everything to cloud storage and offload the processing that makes sense to a cloud service. Exchange does the heavy lifting, manages the singe instance storage database, checks spam blacklists and serves up mail to phones, PCs and Webmail clients alike; Outlook keeps an offline cache of messages, rules and address books. Put them together and you can search last year's email on a plane flight without replicating the other mail server functions on your laptop. Perhaps because the cloud is still in the evangelical phase there are few hybrid solutions around but surely taking advantage of the best of both worlds makes sense?

So, what kind of a disaster would it take for you to think the cloud wasn't the right place for your company to put its IT? -Mary