Cassandra vs ElasticSearch/OpenSearch

[pschichtel]

I'm currently researching Jaeger deployments for production workloads. ElasticSearch and Cassandra, optionally buffered by Kafka seem to be what's recommended for production.

What I'm wondering though: Why would I pick ElasticSearch/OpenSearch over Cassandra or vice versa, while ignoring personal preference?

I have some experience operating ElasticSearch, but I would be able and willing to invest time into Cassandra, if it is the better tool for the job and conceptionally it feels like Cassandra would be a better fit for the use-case as ElasticSearch is most often used as a Full Text Index / Search Engine.

https://www.jaegertracing.io/docs/2.16/faq/#what-is-the-recommended-storage-backend explains that Cassandra has faster writes but limitations around search. Yet people seem to pick Cassandara.

Any pointers to existing work on the topic or experiences you had in the past would be highly appreciated!

[yurishkuro]

https://www.jaegertracing.io/docs/2.16/faq/#what-is-the-recommended-storage-backend

You already showed this link which answers your question:

The Jaeger team recommends Elasticsearch or OpenSearch as the storage backend over Cassandra

[pschichtel]

hmm, on reading the faq entry again I might have interpreted too much into it. I still went for Cassandra for the simpler maintenance and out of personal interest in Cassandra. Depending on how this goes I might still switch it over to ES before going to production with this.