Baseline machine metrics across cluster failovers
CONTEXT: Using the Analysis graph for historical performance analysis of machine metrics on a cluster that failed over at least once
PROBLEM: Machine metrics are collected per node. Baselines of only one node can be shown at a time, and so no baseline view of performance on the "active" component can be synthesised.
EXAMPLE SOLUTIONS:
• Enable multiple simultaneous baselines
• Implement synthetic (bimodal) baselines across cluster nodes
• Provide synthetic metrics for entites on the "active" component of a cluster
What do you think about these solutions? Got another idea? Let us know in the comments!
-
Phil Grayson commented
After failing over a failover cluster, you lose the ability to easily compare the current data with historical data. It would be great if the cluster name was shown as an option and selecting that would only show historical data from the active node.