-
Task
-
Resolution: Unresolved
-
Major
-
None
-
None
I think the number that'd be really interesting is: "how much end-to-end latency gets added via an SMT?", as represented by (kafka message timestamp - source.ts_ms).
One could e.g. do 1M data changes with and without SMT and evaluate the numbers using a simple consumer. Three scenarios to compare:
- No SMT
- No-op SMT
- Event flattenting SMT
- relates to
-
DBZ-1929 Add performance test for SMTs like filters
- Closed