Cluster performance metrics

Cluster metrics monitor cluster performance at a high level. Cluster metrics are aggregated across all nodes in the cluster. OpsCenter tracks a number of cluster-wide metrics for read performance, write performance, memory, and capacity. Watching for variations in cluster performance can signal potential performance issues that might require further investigation. For general performance monitoring, watch for spikes in read and write latency, along with an accumulation of pending operations. Drilling down on high-demand tables can further pinpoint the source of performance issues with an application.

Data Size [data-load]: The live disk space used by all tables on a node.
Heap Used [heap-used]: Average amount of Java heap memory used.
JVM CMS Collection Count [cms-collection-count]: Number of concurrent mark sweep garbage collections performed per second.
JVM CMS Collection Time [cms-collection-time]: Average number of milliseconds spent performing CMS garbage collections per second.
JVM G1 Old Collection Count [g1-old-collection-count]: Number of G1 old generation garbage collections performed per second.
JVM G1 Old Collection Time [g1-old-collection-time]: Average number of milliseconds spent performing G1 old generation garbage collections per second.
JVM G1 Young Collection Count [g1-young-collection-count]: Number of G1 young generation garbage collections performed per second.
JVM G1 Young Collection Time [g1-young-collection-time]: Average number of milliseconds spent performing G1 young generation garbage collections per second.
JVM ParNew Collection Count [par-new-collection-count]: Number of ParNew garbage collections performed per second. ParNew collections pause all work in the JVM but should finish quickly.
JVM ParNew Collection Time [par-new-collection-time]: Average number of milliseconds spent performing ParNew garbage collections per second. ParNew collections pause all work in the JVM but should finish quickly.
Read Requests [read-ops]: The number of read requests per second on the coordinator nodes, analogous to client reads. Monitoring the number of requests over a given time period reveals system read workload and usage patterns.
Read Request Latency (percentiles) [read-histogram]: The min, median, max, 90th, and 99th percentiles of a client reads. The time period starts when a node receives a client read request, and ends when the node responds back to the client. Depending on consistency level and replication factor, this may include the network latency from requesting the data’s replicas.
Total Bytes Compacted [total-bytes-compacted]: Number of bytes compacted per second.
Total Compactions [total-compactions-completed]: Number of sstable scans per second that could result in a compaction.
Write Requests [write-ops]: The number of write requests per second on the coordinator nodes, analogous to client writes. Monitoring the number of requests over a given time period reveals system write workload and usage patterns.
Write Request Latency (percentiles) [write-histogram]: The min, median, max, 90th, and 99th percentiles of a client writes. The time period starts when a node receives a client write request, and ends when the node responds back to the client. Depending on consistency level and replication factor, this may include the network latency from writing to the replicas.