crimson profiling by rados bench metrics
4 KB rados bench test, total 10 round, each round 600s.
single OSD instance
deployed by vstart.sh
crimson-seastore
master
(commit ID bd94f885ab22c9974301f1caca91aea37ecc8edc
)+ #42957
crimson Build configuration
./do_cmake.sh -DWITH_SEASTAR=ON -DWITH_MGR_DASHBOARD_FRONTEND=OFF -DCMAKE_BUILD_TYPE=RelWithDebInfo
[build]$ MGR=0 MON=1 OSD=1 MDS=0 RGW=0 ../src/vstart.sh -n -x --without-dashboard --seastore --crimson --nodaemon --redirect-output
[build]$ bin/ceph osd pool create rbd 128 128 && bin/ceph osd pool set --yes-i-really-mean-it rbd size 1 && bin/ceph osd pool --yes-i-really-mean-it set rbd min_size 1
[build]$ ../../cbt/tools/crimson/seastore_radosbench_run.sh ( set TOTAL_ROUND=10 and BENCH_SECONDS=600)
[build]$ ../../cbt/tools/crimson/seastore_radosbench_analyze.py
...
Transaction invalidate/commit ratio by src
Transaction invalidate/commit ratio for each src, elaborated by the extent-type that caused the conflicts
Transaction invalidated/committed effort ratio for each src
Write amplification at different layers
Write amplification at FRESH_EXTENTS+DELTA layer, elaborated by src, and elaborated by the extent-type that was committed