cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
707
Views
0
Helpful
3
Replies

round robin algorithm - SMTP

jcarvalh
Level 1
Level 1

Hello all.

I am trying to load balance (with round robin) two exchange servers. I am not using stickyness and all session go to the same server.

I search in forums and usually problems like this also involve stickyness.

My config is:

rserver host 10.x.y.z

  ip address 10.x.y.z

  inservice

rserver host 10.x.y.w

ip address 10.x.y.w

inservice

serverfarm host MAIL

  rserver 10.x.y.z

    inservice

  rserver 10.x.y.w

    inservice

class-map match-any MAIL

  100 match virtual-address 192.168.a.b

policy-map type loadbalance first-match MAIL-SF

  class class-default

     serverfarm MAIL

policy-map multi-match VIPS

   class MAIL

      loadbalance vip inservice

      loadbalance policy MAIL-SF

      loadbalance vip icmp-reply

interface vlan 100

   service-policy input VIPS

I think the config is simple and I do not see where is the problem.

Right now I have both rserver operationals and I have 8000 hits on the first one and 0 on the second.

Any help would be appreciated.

Best regards,

Joao.

P.S. - the guys that manage the exchange insist that the performance of exchange servers will increase with round robin (opposed to least conn); but they don't explain why.

3 Replies 3

Daniel Arrondo Ostiz
Cisco Employee
Cisco Employee

Hi Joao,

As you said, your configuration is very simple, so both servers should be getting connections, regardless of the load-balancing algorithm being used.

Could you check the "show serverfarm detail" command to see if you are seeing connection failures on the second server? You are not using any probes, so it could be that, even though the ACE believes the servers is operational, it's actually down and dropping all connections sent to it.

Regards

Daniel

Hello Daniel.

Both servers have a weight of 8 (default values) and the second server is operational, has zero total connection and zero failures.

I manually tried to telnet the second server on port 25 and it worked. Right now I have 20 000 hits in one server and zero on the other.

Thanks,

Joao.

Hi Joao,

That's strange. Probably it would be a good idea opening a TAC service request to have this investigated further.

Daniel

Review Cisco Networking for a $25 gift card