High Availability in the Arctic
My primary purpose for trip was to install a high-availability server cluster running Avance from Stratus Technologies: http://www.stratus.com/products/avance/index.htm. I installed the solution in July when the weather is in the 40s, and the sun shines around the clock. The platform I installed is so dependable that no one should need to visit the site again for a long time. And hopefully no one has to go back in the winter, when the weather can be 50 below and the sun doesn’t come up for months.
To guarantee the uptime we needed, we chose the Avance software in order to have a cost-effective high availability virtual cluster. The solution is so fault tolerant, an entire physical server can fail, and the virtual machines that were running on the failed server will start right back up on the surviving node. It truly separates the running virtual machines from the hardware. Avance monitors the health of the system, all the way down to the status of fans and hard drives. If anything on a node fails, it will automatically migrate the virtual machines off of it and send out a notification.
The Avance software is loaded on two physical servers. That creates a single logical platform to run virtual servers across. Data is continuously replicated between the two nodes. Because of the replication, it doesn’t require a SAN. That not only saves money, but it also reduces the number of moving parts that can fail. Everything, and I mean everything, can be done remotely to maintain the servers. For instance, I can shut down a node and start it back up, I can update firmware, and I can of course setup and teardown virtual servers. All of the management and configuration is done through a well designed and organized web console.
The initial installation called for just two virtual Windows 2003 servers. But now that the platform is in place, additional virtual machines can be brought online at will. The two Dell PowerEdge 2950 servers we installed can easily support a dozen or more virtual servers. Avance is based on XenServer. So now, no matter what server needs the scientists have, we can provide them with servers almost on demand. You can read more about the overall project here: http://blog.timrettig.com/barrow-alaska/