Thanks for this, it makes sense: Ralph Einfeldt wrote:
>mod_jk should be configured to make no balancing. It just should do >the routing. So a1 should route all request with no session info to >t1 and a2 to t2. (Set the lbfactor to 0 for the second worker). Will failover work in this configuration? I guess it won't be able to be setup like this with uneven nos. of Tomcat and apache -----Original Message----- From: Ralph Einfeldt [mailto:[EMAIL PROTECTED] Sent: 27 August 2003 09:03 To: Tomcat Users List; [EMAIL PROTECTED] Subject: RE: mod_jk load balancing with multiple apache servers See below: > -----Original Message----- > From: Peter Anning [mailto:[EMAIL PROTECTED] > Sent: Wednesday, August 27, 2003 11:20 AM > To: Tomcat.Users.List" > <[EMAIL PROTECTED]>"@www.peteranning.com > Subject: RE: mod_jk load balancing with multiple apache servers > > Http request 1 is sent to A1 by the ext LB this request may > go to T1. Then http request 2 is sent to A2 this request may also go to T1. > Now T1 has double the load of T2 which is currently at zero. > > What I am trying to work out in my is: > 1. Does this matter or should it somehow be biased? > 2. Will things be evened out as the no of requests increase? If you can assume an even distribution of the session and the session duration it should be evened out. (It's dependend on the distribution and the quality of the loadbalancer) > 3. How will it scale when adding non-even nos. of Tomcats to Apaches? The number of nodes should not play a role. > 4. How to represent this in mathematical terms? > To express that you have to know the algorithm that is used by the loadbalancer and the distribution of the sessions. BTW: mod_jk should be configured to make no balancing. It just should do the routing. So a1 should route all request with no session info to t1 and a2 to t2. (Set the lbfactor to 0 for the second worker). Note: This recommendation is based on reading rather than on really trying it, havn't used the loadbalaning feature of mod_jk by now. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
