mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-21 00:21:49 +00:00
Latency issue measurements
Wether callbacks are on or not the issue is the same, the first request of a user is expensive, and it gets worse if many users are making their first request at the same time
This commit is contained in:
parent
3f663aea8c
commit
3feee539c5
2 changed files with 130 additions and 0 deletions
65
callbacks_off_baseline.txt
Normal file
65
callbacks_off_baseline.txt
Normal file
|
|
@ -0,0 +1,65 @@
|
|||
=== 42 requests, 42 users, fire-as-fast-as-possible ===
|
||||
Latency = time from request start until last byte of response received
|
||||
[OK] User 19 request 1: 4.144s
|
||||
[OK] User 24 request 1: 3.493s
|
||||
[OK] User 38 request 1: 1.644s
|
||||
[OK] User 4 request 1: 6.345s
|
||||
[OK] User 3 request 1: 6.493s
|
||||
[OK] User 6 request 1: 6.075s
|
||||
[OK] User 12 request 1: 5.247s
|
||||
[OK] User 39 request 1: 1.615s
|
||||
[OK] User 41 request 1: 1.364s
|
||||
[OK] User 20 request 1: 4.183s
|
||||
[OK] User 31 request 1: 2.732s
|
||||
[OK] User 42 request 1: 1.254s
|
||||
[OK] User 1 request 1: 6.857s
|
||||
[OK] User 36 request 1: 2.082s
|
||||
[OK] User 9 request 1: 5.730s
|
||||
[OK] User 23 request 1: 3.834s
|
||||
[OK] User 34 request 1: 2.360s
|
||||
[OK] User 33 request 1: 2.494s
|
||||
[OK] User 35 request 1: 2.227s
|
||||
[OK] User 11 request 1: 5.470s
|
||||
[OK] User 25 request 1: 3.580s
|
||||
[OK] User 21 request 1: 4.117s
|
||||
[OK] User 14 request 1: 5.067s
|
||||
[OK] User 37 request 1: 1.975s
|
||||
[OK] User 18 request 1: 4.522s
|
||||
[OK] User 40 request 1: 1.576s
|
||||
[OK] User 22 request 1: 3.984s
|
||||
[OK] User 13 request 1: 5.208s
|
||||
[OK] User 15 request 1: 4.937s
|
||||
[OK] User 26 request 1: 3.451s
|
||||
[OK] User 8 request 1: 5.906s
|
||||
[OK] User 30 request 1: 2.923s
|
||||
[OK] User 16 request 1: 4.803s
|
||||
[OK] User 17 request 1: 4.672s
|
||||
[OK] User 5 request 1: 6.323s
|
||||
[OK] User 2 request 1: 6.761s
|
||||
[OK] User 10 request 1: 5.635s
|
||||
[OK] User 27 request 1: 3.329s
|
||||
[OK] User 32 request 1: 2.664s
|
||||
[OK] User 29 request 1: 3.065s
|
||||
[OK] User 7 request 1: 6.054s
|
||||
[OK] User 28 request 1: 3.213s
|
||||
|
||||
[OK] 42 succeeded, [FAIL] 0 failed
|
||||
|
||||
Latency (successful):
|
||||
min: 1.254s
|
||||
avg: 4.034s
|
||||
max: 6.857s
|
||||
p95: 6.345s
|
||||
p99: 6.761s
|
||||
|
||||
Requests above threshold (successful only):
|
||||
Above 1s: 42/42 (100.0%)
|
||||
Above 2s: 36/42 (85.7%)
|
||||
Above 3s: 29/42 (69.0%)
|
||||
Above 4s: 21/42 (50.0%)
|
||||
Above 5s: 14/42 (33.3%)
|
||||
Above 6s: 7/42 (16.7%)
|
||||
Above 7s: 0/42 (0.0%)
|
||||
Above 8s: 0/42 (0.0%)
|
||||
Above 9s: 0/42 (0.0%)
|
||||
Above 10s: 0/42 (0.0%)
|
||||
65
callbacks_on_baseline.txt
Normal file
65
callbacks_on_baseline.txt
Normal file
|
|
@ -0,0 +1,65 @@
|
|||
=== 42 requests, 42 users, fire-as-fast-as-possible ===
|
||||
Latency = time from request start until last byte of response received
|
||||
[OK] User 29 request 1: 2.908s
|
||||
[OK] User 9 request 1: 5.687s
|
||||
[OK] User 1 request 1: 6.917s
|
||||
[OK] User 6 request 1: 6.203s
|
||||
[OK] User 37 request 1: 1.901s
|
||||
[OK] User 30 request 1: 2.907s
|
||||
[OK] User 24 request 1: 3.728s
|
||||
[OK] User 27 request 1: 3.315s
|
||||
[OK] User 13 request 1: 5.271s
|
||||
[OK] User 20 request 1: 4.310s
|
||||
[OK] User 33 request 1: 2.524s
|
||||
[OK] User 16 request 1: 4.879s
|
||||
[OK] User 21 request 1: 4.187s
|
||||
[OK] User 15 request 1: 5.039s
|
||||
[OK] User 34 request 1: 2.427s
|
||||
[OK] User 4 request 1: 6.620s
|
||||
[OK] User 17 request 1: 4.806s
|
||||
[OK] User 12 request 1: 5.502s
|
||||
[OK] User 19 request 1: 4.533s
|
||||
[OK] User 10 request 1: 5.799s
|
||||
[OK] User 18 request 1: 4.686s
|
||||
[OK] User 2 request 1: 6.928s
|
||||
[OK] User 28 request 1: 3.304s
|
||||
[OK] User 31 request 1: 2.909s
|
||||
[OK] User 8 request 1: 6.099s
|
||||
[OK] User 25 request 1: 3.730s
|
||||
[OK] User 42 request 1: 1.367s
|
||||
[OK] User 39 request 1: 1.795s
|
||||
[OK] User 5 request 1: 6.533s
|
||||
[OK] User 32 request 1: 2.786s
|
||||
[OK] User 26 request 1: 3.628s
|
||||
[OK] User 3 request 1: 6.839s
|
||||
[OK] User 7 request 1: 6.277s
|
||||
[OK] User 11 request 1: 5.723s
|
||||
[OK] User 35 request 1: 2.396s
|
||||
[OK] User 41 request 1: 1.551s
|
||||
[OK] User 22 request 1: 4.204s
|
||||
[OK] User 14 request 1: 5.325s
|
||||
[OK] User 36 request 1: 2.279s
|
||||
[OK] User 40 request 1: 1.715s
|
||||
[OK] User 38 request 1: 2.001s
|
||||
[OK] User 23 request 1: 4.138s
|
||||
|
||||
[OK] 42 succeeded, [FAIL] 0 failed
|
||||
|
||||
Latency (successful):
|
||||
min: 1.367s
|
||||
avg: 4.183s
|
||||
max: 6.928s
|
||||
p95: 6.620s
|
||||
p99: 6.917s
|
||||
|
||||
Requests above threshold (successful only):
|
||||
Above 1s: 42/42 (100.0%)
|
||||
Above 2s: 37/42 (88.1%)
|
||||
Above 3s: 28/42 (66.7%)
|
||||
Above 4s: 23/42 (54.8%)
|
||||
Above 5s: 15/42 (35.7%)
|
||||
Above 6s: 8/42 (19.0%)
|
||||
Above 7s: 0/42 (0.0%)
|
||||
Above 8s: 0/42 (0.0%)
|
||||
Above 9s: 0/42 (0.0%)
|
||||
Above 10s: 0/42 (0.0%)
|
||||
Loading…
Add table
Reference in a new issue