Database Reference
In-Depth Information
In an ideal situation, when all cluster nodes are up and running, multiple SQL Server clusters would not affect
the performance of each other. Each SQL Server cluster instance is running on a separate node. Unfortunately, the
situation becomes much more complex when one of the servers becomes unavailable, and the SQL Server Instance
failed over to another node, as shown in Figure
31-3
. Both SQL Server cluster instances are running on the same
server, competing for CPU and Memory, and affecting each other's performance.
Figure 31-3.
Two-node multi-instance cluster: One node failure
One of the typical approaches to reduce possible performance implications in the case of a failover in a
multi-instance cluster is by building a cluster configuration that reserves some nodes to pick up the load in case of
a failover. With such an approach, a cluster with multiple active instances would have one or more reserved passive
nodes. If one of the active nodes failed, the instance from that node could failover to the reserved, formerly passive
node, without affecting the performance of the other SQL Server cluster instances. Figure
31-4
shows an example of a
two-instance cluster with one reserved passive node.
