WangzJi opened a new issue, #8039:
URL: https://github.com/apache/incubator-seata/issues/8039
Check Ahead
- [x] I have searched the issues of this repository and believe that this is
not a duplicate.
- [ ] I am willing to try to implement this feature myself.
Why you need it?
The current `seata-benchmark-cli` already supports `--mode SAGA`, and when
`--branches > 0` it uses Seata's state machine engine. This is a good first
step, but the current benchmark workload is still relatively lightweight and
benchmark-oriented.
Today, the benchmark can measure the state machine path itself, but it is
still not ideal for evaluating more realistic Saga behavior under
business-style workloads, especially when we want to compare results with AT /
TCC / XA. In particular, the current implementation does not yet provide enough
workload realism, reproducible failure modeling, or richer Saga-specific result
reporting.
So the need here is not to add a brand new transaction mode, but to make the
existing traditional Saga benchmark more representative and more useful for
performance comparison and regression tracking.
How it could be?
This issue is about the existing traditional Saga path only, i.e.
`BranchType.SAGA` and the current `--mode SAGA` support in
`seata-benchmark-cli`.
A possible direction is:
- keep the current state machine engine based implementation
- enhance `SagaModeExecutor` instead of introducing another Saga mode
- make the workload more realistic, for example by supporting optional
DB-backed business actions or more realistic order / inventory / payment style
steps
- add reproducible failure injection, such as configured failure rate,
timeout simulation, or per-step failure control
- improve result reporting so the benchmark output can clearly show
committed / compensated / failed / compensation-failed results
- define clearer semantics for `--branches` in Saga mode, or allow selecting
different predefined state machine shapes while preserving backward
compatibility
- update README with runnable examples and explanation of the Saga-specific
parameters
Expected outcome:
- existing `--mode SAGA` remains backward compatible
- benchmark results become more representative of a traditional
state-machine Saga workload
- compensation scenarios can be triggered and observed more clearly
- final benchmark report becomes more actionable for comparison and diagnosis
Other related information
Current implementation evidence:
-
`test-suite/seata-benchmark-cli/src/main/java/org/apache/seata/benchmark/BenchmarkRunner.java`
- currently only dispatches `AT`, `TCC`, and `SAGA`
-
`test-suite/seata-benchmark-cli/src/main/java/org/apache/seata/benchmark/config/BenchmarkConfig.java`
- currently validates only `AT`, `TCC`, and `SAGA`
-
`test-suite/seata-benchmark-cli/src/main/java/org/apache/seata/benchmark/executor/SagaModeExecutor.java`
- `branches == 0`: simplified mock mode
- `branches > 0`: state machine engine mode
- `test-suite/seata-benchmark-cli/README.md`
- documents SAGA support and state machine usage
Non-goal for this issue:
- annotation-based Saga support (`BranchType.SAGA_ANNOTATION`) should be
tracked separately because it is a different execution path and likely needs
different bootstrap/integration logic.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
