When providing platform hosting, Vail calculates the number of servers needed at peak capacity and then configures twice that number, with all scattered across multiple data centers. At each data center, the servers for the application are divided into at least two computing cells. Thus, if machine server failure does occur, it can simply be taken off-line, and the redundant servers within the cell will pick up the load. In the highly unlikely event that an entire cell fails, the call load is distributed among the three remaining cells, turning what might have been an emergency into standard operating procedure. Only in the event of an entire data center failing do we have anything resembling an emergency.
This careful policy allows Vail to take a single cell off-line for maintenance while still providing full application redundancy. At no time will a single server failure, or a cell maintenance operation, disrupt the redundancy of any application.
Transforming possible emergencies into standard procedures is the reason for Vail's history of excellent customer service. We are always alert to the possibility of something going wrong, and are seldom disappointed-but we ensure that our customers are seldom disappointed either.
