Tuesday 17 August 2010

Cloud Computing - On The Rocks?


Having just found a place in the UK which has neither broadband nor cable nor 3G connections to the internet, it set me thinking ...

Firstly what a great place for a holiday, literally getting away from it all!

Secondly not the best place to run an internet business. Nor to rely on the internet for staff or any form of mobile services when people are in the area - and there are plenty more places without fast internet around the wilder parts of the UK.

Thirdly what's it like when a business loses internet and email access for a week, a fortnight, a month? I've previously looked at ways of avoiding dependency on one method of internet access, and the availability of multi-user MiFi access devices. After all, there are several cases of BT lines being out of action for 10 days or more, as I experienced first-hand.

Overcoming Cloud Provider Problems

But what if there's a problem with the provider at the other end? One commentator wrote a few weeks ago "There has NEVER been a documented case of catastrophic data loss with a cloud service." The comment came across as "it hasn't happened, so it won't happen". Nieve? Whilst cloud providers may have always (up to now) been able to recover from backups in the event of data loss due to internal malfunction, there's going to be a first time. After all, there;s the old saying "Past Success Does Not Guarantee Future Performance". And what about a cloud provider going out of business?

There was after all the tiny matter of the demise of Exodus Communications which entered Chapter 11 bankruptcy in 2001. At one point over half of all internet traffic went through its systems, serving as it did he likes of Google, eBay, Yahoo!, PayPal, BestBuy, Weather Channel, Merrill Lynch, American Airlines, Microsoft, Hotmail, Virgin Mobile and O2 Plc from some 46 data centres (sorry centers). To be fair it was bought a couple of months later with probably little or no data loss to its customers. But it could have been much different.

It also highlights that outsourcing IT systems to cloud providers doesn't involve just one company. Word reaches me that a reseller has gone out of business, and the hosting company behind it has shut off access by the end-users to their systems and data. I wonder how many of those customers even knew who the hosting company was, let alone checked them out in advance?

So what can be done? There are three key steps forward:
  1. "Due diligence" on each provider in the supply chain for each cloud service. Here are some suggestions from a cloud advocate. These need to be continually applied as the supply chain changes.
  2. Making sure you have up-to-date copies of your data and a contingency plan
  3. Avoiding undue reliance on one provider for each application area.
Duplicating systems such as CRM or ecommerce may be rather tricky, time-consuming and could be significantly more expensive. But is it possible at a reasonable cost to have one part of the business on one system, and another part on another system? Then it is possible to have a plan to switch the whole business onto one system in the event of any problems with the other.
What price a good night's sleep?

In any case do comment if you know of any real life examples of cloud data being lost- whether due to bankruptcy, system failure, security breaches or any other reason.

.

1 comment:

  1. About a year ago there was an incidence of data loss "Microsoft Red-Faced After Massive Sidekick Data Loss"
    http://www.pcworld.com/article/173470/microsoft_redfaced_after_massive_sidekick_data_loss.html

    This shows that cloud SaaS providers cannot be automatically trusted to be applying rigorous disaster recovery procedures - even large companies like Microsoft. Indeed the better cloud SaaS vendors provide facilities for data extraction, and encourage their users to use them regularly for belt and braces.

    ReplyDelete