Cluster-level metrics provide aggregated information across all the nodes and is the first place one should look when monitoring an Elasticsearch cluster. Cluster-level metrics are displayed in the Overview tab and can be navigated to by clicking on the Overview link under the Elasticsearch section found in the landing page of the Monitoring UI.

The Overview tab provides key metrics that indicate the overall health of an Elasticsearch cluster:

The key metrics that are displayed are cluster status, number of nodes and number of indices present, memory used, total number of shards present, total number of unassigned shards, total number of documents present in the indices, the disk space used for storing the documents, uptime, and version of Elasticsearch. The Overview tab also displays charts that show the search and indexing performance over time, while the table at the bottom shows information about any shards that are being recovered.

In the Overview tab, the metrics are aggregated at the cluster level; so, when you're monitoring the Elasticsearch cluster, you might miss out some vital parameters that may eventually affect the cluster's overall state. For example, the Memory Used metric showcases the average memory used by combining the memory used across all nodes. However, one node might be running with full memory utilization and another node's memory might have hardly been used. Hence, as an administrator, you should always monitor at the Node level too.