Change Data Capture and Impact on Database Processing
The question is simple, the answer, not so much. There are, however, a number of considerations.
Note SAP HANA is an exception that supports no supplemental logging at all, and HVR implemented a separate approach to perform CDC from this database.
The addition of supplemental logging will impact primary database processing because it results in more transaction log being generated. The amount of extra impact on the database depends on:
To ensure uninterrupted, low-latency CDC, capture must be running faster than the database is writing the logs. HVR’s direct read and pipelined execution ensures optimum efficiency to keep up with the database log writers. As a result, when capture runs continuously, it will be capturing from the tail end of the log where the log writer(s) are writing. While clients occasionally raise a concern for contention on the log in practice there is none, because the busy tail of the log is generally cached, either on the file system, or in the case of Oracle ASM in ASM instance buffers.
HVR CDC processing takes up CPU, memory and IO resources. HVR will always retrieve the changes in system commit order irrespective of whether the database is a single instance or a clustered database, with one reader per thread, and for a clustered database there is a coordinator to ensure changes are put in order. Per reader HVR will at most use one CPU core when running behind, which in practice, on a modern, adequately-sized database server, means there is very limited if any noticeable impact on database processing due to this extra resource consumption.
Interested in learning more? Join one of our live demo webinars.