[Performance] 🟢 MTTFB 38% improved by - Flux.log removal POST /api/login

[ghkdqhrbals]

POST /api/login

• Implements
  • caching FriendsList DTO into Redis
• Resource consumption
  • chat-db : 0.23 cpu
  • chat-server : 0.69 cpu
• Improvements
  • avg MTTFB : 108.5 -> 79.2 👍 38% improvement
  • avg TPS : 16.6 -> 15.6 👎 7% worse - no big difference

Average of MTTFB(Mean Test Time of First Byte) is 38% improved by removing .log(). And TPS isn't that big difference between before and after. Small changes may be affected by things like thread initialization.

This stress test started with 99 concurrent users.

Test during 3 min

Before

Total Vusers	99
Agent	1
Processes Threads	3 / 33
Sample Ignore	0
TPS	16.6
Peak TPS	24
Mean Test Time	5,845.54
Executed Tests	2,891
Successful Tests	2,891
Errors	0

• AVG MTTFB : 108ms

After

Total Vusers	99
Agent	1
ProcessesThreads	3 / 33
Sample Ignore	0
TPS	15.6
Peak TPS	21
Mean Test Time	6,221.57
Executed Tests	2,714
Successful Tests	2,714
Errors	0

• AVG MTTFB : 79ms

[ghkdqhrbals]

After

This stress test started with 1000 concurrent users.

But server dead 😂

Test during 3 min

Total Vusers	1,000
Agent	1
Processes Threads	10 / 100
Sample Ignore	0
TPS	10.9
Peak TPS	22
Mean Test Time	40,830.90 ms
Executed Tests	2,227
Success Tests	1,888
Errors	339 ( This means that server cannot handle with 1,000 concurrent users )