It is great enthusing about the benefits of Cloud Computing, but what are the consequences when it goes wrong ? Of course, there are different levels of ‘going wrong’. We have often publicised outages from the likes of Amazon and Google, but given the publicised SLA’s of each some down time is expected. However things can get much more serious than this. In the last couple of days it has emerged that Carbonite had lost some of its customers data in 2007. According to TechCrunch Carbonite lost the data of 7,500+ customers who relied on the company to keep their files safe. This emerged because of a law suit that Carbonite filed on the providers of their infrastructure. The Carbonite CEO provides more details:
“The failures of the Promise equipment occurred primarily during 2007. We stopped buying the Promise servers and switched suppliers. We allege that the Promise servers had defective firmware and were not reliable enough for Carbonite’s use. We are demanding that Promise compensate us for the cost of replacing their defective products. As for the 7,500 affected customers, their backups were restarted automatically and immediately on our new servers.”
In this case it seems the data loss had no effect and any data loss was mitigated by the companies internal backup procedures. However, in the case of JournalSpace, this unfortunately was not the case. JournalSpace was a blogging platform that had been around for about 6 years and, due to a disgruntled employee, all customers blogs were wiped out from their internal servers. Ouch !
This is not the first time such human error has lead to such problems. In August 2008 Cloud Platform FlexiScale, had an outage for over 2 days due to an engineer accidentally deleting a main storage volume.
Also the damage done it not just to the customer but also to the vendor’s reputation, especially if they are a smaller vendor trying to make a name for themselves. One such catastrophe can literally be the difference between success and failure in the market.
So what can we learn from this ? Well the first thing is that, just like Murphy’s law, if something can go wrong at some point it invariably will. With that in mind you should always take all steps to protect your applications and data. This could mean backing them up locally or keeping backups on different storage clouds and having a DR ready plan in place. If you don’t then you cannot just blame the cloud….