IP, Technology & Data

We’ve blogged on several occasions about the importance of business continuity and disaster recovery plans – most notably about the impact on global supply chains following the giant ash cloud caused by the eruption of Eyjafjallajokull in 2010.

I was surprised at the impact that Hurricane Sandy had on the Internet earlier this week, with several major websites (including the Huffington Post, Gizmodo and Gawker) being knocked offline for several hours as a result of flooding and damage on the east coast of the United States. According to reports, a data centre lost power as a result of a battery failure caused by flooding.

Using third party data centres and IaaS vendors can provide a way of mitigating some of the risks of business interruption. Cloud providers are often better placed to manage these risks as they often operate multiple data centres which should, in theory, mitigate the impact of a single event. But in this case, for whatever reason, that hasn’t happened.

For the websites affected by Hurricane Sandy, that downtime will have led to a substantial loss of advertising revenue across the globe. Assuming that the hosting company has otherwise complied with its contractual obligations, it’s unlikely that the website operators would be able to recover any costs from the hosting company as the hosting company will likely claim relief under the force majeure provisions in its contract.

Geographic separation
When assessing your business continuity arrangements (and those of your suppliers), it’s therefore important that one of the things that you review the proximity of any back-up facility to the primary site. Events such as tropical storms, earthquakes, power failures and civil unrest can affect a large area, meaning that multiple data centres on either side of a single conurbation could easily be affected.

As one person I follow on Twitter said, what will happen to the Internet when the Big One finally hits California?

And it’s not just the data centre location that you need to think about.

I once heard a tale about an organisation who used two telecoms companies to provide physical telecoms links between its primary office and its data centre elsewhere in the city. All was well until roadworks took place on a bridge over a river. Whilst each company used physically separate cables between the premises (including separate points of exit and entry), it turned out that the bridge in question was a single point of failure – both companies had chosen to route their cables across the river using the same bridge – and a single jack hammer blow took out both links.

Follow me