Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

ACE roundrobin architecture


My customer been used used the alteon L4

at these day, We changed from alteon to ACE,

the conversion was successful, but there are some problem in roundrobin loadbalancing

for example, we tried LB test only one host,

the host are trying to connect vip, the time interval was about 1 seconds.

the serverfarm's LB was roundrobin.

the connections likes this

rserver(1)=11', rserver(2)=22'

we supposed that the connection rserver(1) , and rserver(2), and rserver(1), rserver(2),, but there are not.

Is there a difference compare with alteon roundroubin althorism?

anybody who can explane it to me?


Cisco Employee

Re: ACE roundrobin architecture

You would be very lucky to see perfect loadbalancing numbers with just one host.

The roundrobin algorithm is the same but the architecture is different so it gives different results.

The ACE module is actually 2 devices (IXP) inside 1 box. Each IXP will get half of the traffic and perform the loadbalancing decision independently from the other IXP.

The traffic is split between the IXP based on TCP ports or ip addresses or mac addresses.

So, with a simple example of 1 host sending 2 connections, you could see the first connection going to IXP1, IXP1 selecting server 1, the 2nd connetion going to IXP2 also selecting server 1.

So globally both connections went to server 1 and not connection to server 2.

This sounds bad.

But if you take a real life test, thousands of hosts, and millions of connections, you should normally end with more or less equal share between both servers.