Cannot distribute traffics evenly by using round robin

Hello All,

Customer use the round-robin method without any persist template.
Total sessions are not load balancing evenly below;
Surely I checked again after clearing sessions. And the monitor of real servers was normal.
But sessions still distributed evenly.
I think this issue is limited to my customer......:)
However...I wonder if anyone have this experience.


*v112_158_gr_80 State: All Up
r112_143:14080 20 5947332 52365756 51882310 892
r112_144:14080 21 5948479 52238493 51658280 906
r112_145:14080 21 5948154 52180299 51569971 946
r112_146:14080 23 5948650 52240530 49853274 1194

FrontB-L4-1-Active#sh slb server | be r112_14
r112_143:14080/tcp 22 136903414 936764409 1203320848 1148 Up
r112_143: Total 22 136903414 936764409 1203320848 1148 Up
r112_144:14080/tcp 17 136903086 933975503 1193917435 1167 Up
r112_144: Total 17 136903086 933975503 1193917435 1167 Up
r112_145:14080/tcp 21 137015956 934610406 1194983543 948 Up
r112_145: Total 21 137015956 934610406 1194983543 948 Up
r112_146:14080/tcp 24 137013613 934139889 1184469441 1207 Up
r112_146: Total 24 137013613 934139889 1184469441 1207 Up


Thanks,
CB.

Comments

  • dshindshin Member
    edited February 2016
    What ACOS build have you installed? I would like to replicate on our lab? Also what application? Microsoft, standard IIS.....please provide details.

    Genard
  • escgateescgate Member ✭✭
    edited February 2016
    Genard,
    Thank you for replying.

    First, I think I need testing with some verions and IIS web severs.
    I will update the results.

    Thanks,
    CB
  • dshindshin Member
    edited February 2016
    So for round robin algorithm the CPU is monitored during server selection. If a request landed on a CPU that has a heavy load, the request may have to be redirected to another CPU causing the round robin algorithm to not distribute evenly. The A10 round robin has a proprietary selection algorithm based on CPU load. If you want an evenly distributed selection across servers, I would suggest that you try the "round robin strict". The "round robin strict" will offer an equal distribution of request across servers with no consideration of CPU threshold.
Sign In or Register to comment.