[mapguide-users] Strange behavior with Load Balancing

[Bruno Scott]

I have setup a 3 server architecture for testing LoadBalancing with Mapguide
Server 1 : IIS + WebServerExtention
Server 2 : Mapguide server a
Server 3 : Mapguide server b

I have 20 concurent users driven by a grinder script that does this series
of instructions over and over
  Load the viewer + wait 2 seconds
  Loop 20 times
    Zoom in + wait 2 seconds
    Select + wait 2 seconds
    Unselect + wait 2 seconds
    Pan 3 times + wait 2 seconds
    Zoom out + wait 2 seconds
A complete cycle takes about 3 minutes

Here what i got after the first run ( request seems to be evenly balanced )
http://n2.nabble.com/file/n3294688/20users_2MGE_1IIS_stable.jpg 

After the second run ( mg server a gets much more request )
http://n2.nabble.com/file/n3294688/20users_2MGE_1IIS_20_40.jpg 

After the third run  ( mg server a gets almost all request )
http://n2.nabble.com/file/n3294688/20users_2MGE_1IIS_40_60.jpg 

As far as i know the load balancing is suppose to be quite simple :
  Each session is assigned to a server
  And subsequent request using that session is send to that same server
  
  The first Create Session should go to server A
  The second to server B
  The third to server A
  and so on

Do somebody has experienced the load balancing?
Could it be too much sessions created in to short period of time?

Bruno Scott


-- 
View this message in context: http://n2.nabble.com/Strange-behavior-with-Load-Balancing-tp3294688p3294688.html
Sent from the MapGuide Users mailing list archive at Nabble.com.