-
Notifications
You must be signed in to change notification settings - Fork 0
System Tests
Neil Moore edited this page Dec 8, 2015
·
3 revisions
This where the testing plan for this project is detailed.
- Microbenchmarks of hotspots [ NOT IMPLEMENTED ]
- Heuristic accuracy analysis [ NOT IMPLEMENTED ]
- Ability to handle large amounts of network traffic [ NOT IMPLEMENTED ]
- Many connections (both short-lived and long-lived)
- Many packets
- Both
- Tokenizer / parser accuracy test (regression tests) [ NOT IMPLEMENTED ]
- Parser speed benchmark [ NOT IMPLEMENTED ]
- Query plan accuracy test (regression tests) [ NOT IMPLEMENTED ]
- Query execution speed benchmark [ NOT IMPLEMENTED ]
- Access speed benchmark (regression tests) [ NOT IMPLEMENTED ]
- Query evaluation speed benchmark [ NOT IMPLEMENTED ]
- Throughput on contested tables [ NOT IMPLEMENTED ]
- Many readers and writers on a single table
- This tests Tervel's ability to manage many threads
- Cache coherency/efficiency benchmark [ NOT IMPLEMENTED ]
- This can be measured using cachegrind
- Data consistency (regression tests) [ NOT IMPLEMENTED ]
- Overall system load [ NOT IMPLEMENTED ]
- Many connections, many queries, large tables, many tables
- Accuracy tests [ NOT IMPLEMENTED ]
- Similar to SQLLogicTest
- Contested resources [ NOT IMPLEMENTED ]
- Pegged threads in background doing I/O, lower than usual memory capacity
- Continuous operation [ NOT IMPLEMENTED ]
- How does the system handle running at load for a long time?
- End-to-End latency benchmark [ NOT IMPLEMENTED ]
- Varying system loads, how does latency change with the load on the system?