0

Seeing high latency for range queries with ALLOW FILTERING and CL LOCAL_QUORUM in Cassandra 4.0.11.

I see repaired_data_tracking_for_range_reads_enabled property in cassandra.yaml which can impact the range query performance. I have set it to false. Also there is replica_filtering_protection which is kind of guardrail and in my opinion should not have impact on performance.

Is there anything else introduce in Cassandra 4 with respect to range reads that needs to be adjusted? I am comparing nodetool proxyhistograms output of Cassandra 3 and Cassandra 4 for same load.

1 Answer 1

1

There is insufficient detail in your post to be able to provide any meaningful answer other than I am not aware of any specific changes introduced in Cassandra 4.0 that would lead to a reduction in performance of range reads.

As a general rule, comparisons between major versions is often problematic since they are usually not apples-for-apples. Having said that, it is reasonable to expect better performance from the newer version or at the very least the same level of performance provided the comparison is like-for-like.

You will need to conduct additional investigation on your end to try and isolate the source of the latency if there is indeed an issue. At the very least, you should trace the query to identify where Cassandra is spending most of its time to generate the result.

Once you have something concrete, you will need to document the steps on how to replicate the problem including crucial details such as the table schema, problematic query(s), and sample data. Cheers!

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.