Cassandra Monitoring and Performance Management
Cassandra Monitoring is part of Instana’s AI Powered Microservice APM solution. Apache Cassandra is an open source distributed NoSQL database designed to handle large amounts of data across a number of servers with no single point of failure. It specializes in supporting clusters than span across many data centers, and to this end is capable of asynchronous masterless replication. This is a highly flexible methodology, but it does come with a cost. A common downside is that Cassandra is not row level consistent; inserts and updates on the same table that effect the same row may clash, resulting in inconsistent data.
If Instana is deployed into an infrastructure containing Apache Cassandra, the agent will automatically detect the technology and deploy the correct sensor to monitor Cassandra and its interaction with the database. No human setup or configuration is required.
Cassandra Performance and Configuration Monitoring
After Instana automatically deploys its Cassandra sensor, it will immediately map out Cassandra’s infrastructure: the individual write-back cache nodes in relation to their respective datacenters, and those datacenters in relation to their clusters. Instana collects pertinent performance metrics at both the node and cluster levels to determine the interaction quality between the Cassandra nodes and clusters. The Instana agent will send those metrics back to our Dynamic Graph model, which stores and contextualizes all collected monitoring data. Typical metrics collected at the node level are:
- Client Read/Write Latencies
- Pending Requests
Some example metrics collected at the cluster level are:
- Keyspace Details
- Replication Factors
- Disk Sizes
Please see the Instana Cassandra Management Documentation for a complete list. Based on their infrastructure, developers can configure the threshold limits for Instana to alert them with context around the point of failure.
Cassandra Monitoring – Health
Instana will detect configuration changes and tack them on the timeline, for example changes to cluster configuration and the addition or removal of key spaces.