Learn F5 Technologies, Get Answers & Share Community Solutions Join DevCentral

Questions and Answers


Loading

Submitting

Understanding Round Robin

question
Hoping someone can clarify this for me. I have several VIP's that all use the same pool. This pool uses the "Round Robin" load balancing method. Is each connection to each VIP round robin'd independently of each other? So, all connections to VIPA are round robin'd, and all connections to VIPB are round robin'd amongst all members in the pool. or.. connection to VIP1 hits webserver1, then connection to VIP2 hits webserver2? Hope this makes sense. Thanks, Josh

3 Answer(s):

What you have is correct. Round Robin is handled on a per Virtual Server basis, even if there is another Virtual Server using the same servers, or even the same pool.

Using Round Robin does not keep the load distribution equal to all of your servers (It doesn't care. If it is a servers turn in the rotation, it gets the traffic regardless of the current load).

If you wish to keep your connections equal per Pool, use "Least Connections (member)". - This tracks all of the connections for each server, for this pool, on this Virtual Server. (If the same pool is being used by a different Virtual Server those connections are handled seperately).

If you wish to keep your connections equal per Server, use "Least Connections (node)". - This tracks all of the connections to the Node in all Pools.

Ok, hrm. I just did some brief testing, and it looks like that the VIP's are not being round robin'ed on a per VIP basis. It looks like that all all connections to all VIPs are in the same "round robin pool" ie..

- Connection comes into VIP1 - hits webserver1
- Connection comes into VIP2 - hits webserver2

not:

- Connection comes into VIP1 - hits ws1
- 2nd connection comes into VIP1 - hits ws2

- Connection comes into VIP2 - hits ws1
- 2nd Connection comes into VIP2 - hits ws2

This make sense?
Makes sense. I was under the impression that they were handled independently for each VIP.

I guess that it is handled on a per Pool Basis.

Your answer: