Which use case/requirement will be addressed by the proposed feature?

Recently, we configured the Cassandra connector and are experiencing poor performance when reading the commit log files. After reviewing the source code and conducting some tests and profiling, I determined that the call to pu.maxTimestamp() within the XCommitLogReadHandlerImpl classes is taking up excessive time and significantly slowing down the connector.

Implementation ideas (optional)

I recommend evaluating pu.maxTimestamp() once at the start of the loop in handleRowIterator and then passing the value to the handleRowModifications method. I tested this idea with the Cassandra 3 connector and observed no alterations in the logic, while the performance improved significantly.

I would be pleased to submit a PR on a GitHub repository.