Picked up some good experience recently in setting up a BizTalk environment with no single points of failure. That’s two BizTalk Servers in a group running off an active/passive clustered SQL back end with the master secret server running on the SQL Cluster. It was all very easy. I set up the cluster first, then installed SQL and service packs across the cluster, done a custom BizTalk install only selecting the Enterprise SSO components to be installed on each cluster node, ran the BizTalk configuration wizard to create the SSO group on the cluster, clustered the SSO Service as a generic service, backed up the master secret to the passive server and tested fail over, then I installed BizTalk on the BizTalk Servers, ran the configuration wizard pointing to the SQL Cluster and joining the SSO group on the Cluster, then joined the second BizTalk Server to the group.
It works well. We deploy solutions in symmetry across the BizTalk Servers so that we can pull any BizTalk Server out the group for maintenance without impacting availability. In the rare event that a SQL Server has a hardware meltdown the active cluster node fails over with the BizTalk host instances automatically recycling resulting in an outage of approximately a minute. If we loose the SSO Service on the cluster it automatically fails over to the passive node. Oh yeah and storage was on a SAN with the disk resource being clustered in the same group as the SQL Cluster. Too easy!
Anyhow, was good turning theory into practice and getting some hands on experience.