Over the past few months we have been upgrading our infrastructure to provide improved redundancy to our customers. Specifically we’ve enhanced our geographic redundancy so there is now no single point of failure.
We’ve always had redundancy – so that if a server fails then there will be no interruption to the service of our clients. This is absolutely necessary because servers can fail – particularly disks. However we used to be vulnerable to a whole data center going off line. The data centers we use go to a lot of effort to ensure that they don’t go off line – they have multiple internet connections, a lot of redundant power systems, and lots of security. However even if they guarantee 100% up time – it doesn’t mean that you always get 100%. And despite all of these efforts we have had our data centers go off line for short periods. When this happened our customers experienced a temporary loss of service (while we execute our disaster recovery plan) and even a small loss of service is unacceptable to us so we have been looking for a solution. This upgrade is a major part of the solution.
The technique we use to achieve the geographic redundancy is the same as that used by many of the big players, like Google and Microsoft- but as far as we can tell not by our competitors – like Mercado (what’s left of them now belongs to Omniture), Atomz (now run by Omniture), ThanxMedia, or Celebros. Now if one of our data centers goes off line then the others will take the load and we will continue providing service to our customers.
This change is transparent to our customers and costs them nothing extra. It is part of our effort to continuously improve our service. Web site owners need to know that their search and navigation is going to be running continuously. Last week Evan Schuman predicted that e-commerce site crashes will soar this holiday season. This upgrade will help ensure that our customers’ sites will function continuously.