Load testing Ayase + Torako. Each run represents the best output with no dropped requests.
- https://archive.org/details/archivedmoe_db_201908
- /3/, /a/, /aco/, /adv/, /an/, /asp/, /b/, /bant/, /biz/, /can/, /cgl/, /ck/, /cm/, /cock/, /con/, /c/
- ~500M rows
- https://github.com/miyachan/torako
- Config:
- https://gist.github.com/miyachan/4d214b6f0823864cfa51dd25951d3212
- Scraping every board
- Use of Asagi Triggers disabled (sister tables are managed by Torako internally without triggers).
- https://gist.github.com/miyachan/4d214b6f0823864cfa51dd25951d3212
- Google Compute Engine
- e2-highcpu-16
- 16 vCPU
- 16GiB memory
- nginx Frontend
worker_connections = 8192;
worker_rlimit_nofile 30000;
- open file limit = 33554432
- Ayase
uvicorn --workers $(($(nproc) * 3 + 1)) --log-level error --no-access-log --loop uvloop --backlog 65536 view.asagi:app
- Foolfuuka
- fpm:
listen.backlog = 65535
- fpm:
pm.max_children = 128
- fpm:
- Google Cloud SQL Database
- 2 vCPU
- 7.5 GiB memory
- Google Compute Engine
- e2-highcpu-16
- 16 vCPU
- 16 GiB memory
urls.conf
GET http://10.128.0.5/a
GET http://10.128.0.5/3
GET http://10.128.0.5/aco
GET http://10.128.0.5/adv
GET http://10.128.0.5/an
GET http://10.128.0.5/asp
GET http://10.128.0.5/b
GET http://10.128.0.5/bant
GET http://10.128.0.5/biz
GET http://10.128.0.5/cgl
GET http://10.128.0.5/ck
GET http://10.128.0.5/cm
GET http://10.128.0.5/c
GET http://10.128.0.5/a/thread/208415801
GET http://10.128.0.5/a/thread/208416511
GET http://10.128.0.5/a/thread/208413548
GET http://10.128.0.5/a/thread/208417410
GET http://10.128.0.5/a/thread/208347709
GET http://10.128.0.5/a/thread/208407848
GET http://10.128.0.5/a/thread/208419443
GET http://10.128.0.5/a/thread/208383789
GET http://10.128.0.5/a/thread/208408998
GET http://10.128.0.5/a/thread/208327181
GET http://10.128.0.5/a/thread/208398803
GET http://10.128.0.5/a/thread/208417410
$ cat urls.conf | ./vegeta attack -duration=120s -rate=150/1s -max-body 4b -header "Host: archive.wakarimasen.co" | tee results.bin | ./vegeta report
Requests [total, rate, throughput] 18000, 150.01, 149.75
Duration [total, attack, wait] 2m0s, 2m0s, 203.118ms
Latencies [min, mean, 50, 90, 95, 99, max] 21.335ms, 1.587s, 963.271ms, 4.113s, 4.302s, 4.468s, 4.644s
Bytes In [total, mean] 72000, 4.00
Bytes Out [total, mean] 0, 0.00
Success [ratio] 100.00%
Status Codes [code:count] 200:18000
$ cat urls.conf | ./vegeta attack -duration=120s -rate=150/1s -max-body 4b -header "Host: archive.wakarimasen.co" | tee results.bin | ./vegeta report
Requests [total, rate, throughput] 18000, 150.01, 76.61
Duration [total, attack, wait] 2m30s, 2m0s, 29.674s
Latencies [min, mean, 50, 90, 95, 99, max] 11.63ms, 14.424s, 9.161s, 30s, 30s, 30.001s, 30.008s
Bytes In [total, mean] 45864, 2.55
Bytes Out [total, mean] 0, 0.00
Success [ratio] 63.70%
Status Codes [code:count] 0:6534 200:11466
FoolFuuka | Ayase | |
---|---|---|
Req/s | 149.7 r/s | 76.61 |
Avg Latency | 1.5s | 14.4s |
Median Latency | 963ms | 9.1s |
99% Latency | 4.4s | 30s |