forked from tarantool/tarantool
-
Notifications
You must be signed in to change notification settings - Fork 0
SQL performance benchmarks
Ilya Petukhov edited this page Nov 30, 2017
·
7 revisions
Start benchmark:
curl -X POST \
-F token={TOKEN} \
-F ref=master \
-F "variables[BRANCH]={BRANCH}" \
-F "variables[TIME]={TIME}" \
https://gitlab.com/api/v4/projects/4294910/trigger/pipeline
Where:
- {TOKEN} is trigger token. You should get it here (see tab "Pipeline triggers")
- {BRANCH} is name of the tarantool branch to be tested. By default, it is 1.8
- {TIME} is the time of the test run in seconds. By default, it is 2400 (i.e. 40 min).
You can see the status of the benchmark run here.
Sysbench consists of 11 tests:
- oltp_read_only
- oltp_point_select
- oltp_insert
- oltp_update_index
- oltp_update_non_index
- select_random_points
- select_random_ranges
- bulk_insert
- oltp_write_only
- oltp_read_write
- oltp_delete
Start benchmark:
curl -X POST \
-F token={TOKEN} \
-F ref=master \
-F "variables[BRANCH]={BRANCH}" \
-F "variables[TIME]={TIME}" \
https://gitlab.com/api/v4/projects/4186534/trigger/pipeline
Where:
- {TOKEN} is trigger token. You should get it here (see tab "Pipeline triggers")
- {BRANCH} is name of the tarantool branch to be tested. By default, it is 1.8
- {TIME} is the time of the test run in seconds of EACH test. By default, it is 220.
You can see the status of the benchmark run here
You can see the results here: http://bench.tarantool.org
Architecture Specifications
- Server architecture
- Feature specifications
- What's in a good specification
- Functional indexes
- Space _index structure
- R tree index quick start and usage
- LuaJIT
- Vinyl
- SQL
- Testing
- Performance
How To ...?
- ... add new fuzzers
- ... build RPM or Deb package using packpack
- ... calculate memory size
- ... debug core dump of stripped tarantool
- ... debug core from different OS
- ... debug Lua state with GDB
- ... generate new bootstrap snapshot
- ... use Address Sanitizer
- ... collect a coredump
Lua modules
Useful links