Jul 2, 2012
Twice in June Amazon's Data Centre in North Virginia was knocked out of action. The first outage was on 15th June and caused by a generator fan failure and the second time was over this weekend due to local weather knocking out the power supply. This took out Netflix, Pinterest, Instagram and other high profile consumer cloud services but has no doubt also affected any smaller businesses relying on Amazon for IT in the Cloud.
This is another reminder that cloud services are subject to outages and local events no matter how many layers of redundancy are put in place. It also highlights that where you are making cost savings by consolidating a large amount of services the time to recover may also increase due to the sheer volume of users sharing the service. In the latest outage power was off for about 30 minutes but many services were not up as at Monday 2nd July simply because it takes time for the Amazon engineers to restart all the virtual servers in the datacentre.
• When planning IT infrastructure, whether in the Cloud or on-premises, you should consider outages.
• Can you work if you have limited or no computing power?
• How fast can you recover and is that within your control?
• Who has your data and what alternatives do you have to get running again?
The key metric is the cost per hour to your business if you do not haveaccess to your software programs and/or your data.
Considering this may give you a better idea of value when looking at options for faster vs slower recovery times.