DSE查询优化和用例

Question

Should I use Solr for all of my reading activities and then Cassandra for all writes to maximise the performance of DSE? or can I read using Cassandra but obviously on a key value basis for select activities?

Answer 1

Cassandra is a write-optimised database and so reads may be slow, but Solr should be used a crutch or a 'nitro-boost' if you will, and not as the go-to method for reading. Because if your reads are slow, it may be because the DB design is fundamentally flawed and that could be dangerous for scaling as well as maintenance.

Maximizing the performance of a DSE should be based on the pattern of your reads and writes. For example if your users table is only used for login and a couple of other times for profile related data, you don't need Solr for that. Some duplicate tables with different keys should suffice.

However if your app is an ERP that requires user data at all times, Solr indexing for faster reads should be considered.

And to reiterate, if your reads are slow, check if a better db design can solve the issue.