ferrarista
asked on
ActiveSync - CAS
I'm not quite sure how ActiveSync works when working with two load balanced CAS servers.
Do they use the same CAS server for a certain lenght of time ? Or do they costantly talk to either server ?
I was doing a test like refreshing my inbox on the mobile device and I can actually see IIS traffic on both CAS servers. So how does it actually work ? I cannot find any material to explain this. Would love to get an insight.
Thanks so much.
Do they use the same CAS server for a certain lenght of time ? Or do they costantly talk to either server ?
I was doing a test like refreshing my inbox on the mobile device and I can actually see IIS traffic on both CAS servers. So how does it actually work ? I cannot find any material to explain this. Would love to get an insight.
Thanks so much.
Akhater
Depends on your load balancing technology
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
We are actually using round robin.
What puzzles me is that I find transactions on both servers at basically the same time, so that got me thinking....
I'm actually trying to resolve a problem where EAS performance has suddenly decreased.
What puzzles me is that I find transactions on both servers at basically the same time, so that got me thinking....
I'm actually trying to resolve a problem where EAS performance has suddenly decreased.
No a connection will happen only to one of the servers at a time
Round Robin isn't load balancing.
If you are seeing problems, then take it out. Ensure all traffic goes to one server.
If you want HA for CAS, then you need a load balancer. Kemp or JetNexus are the most common ones I deploy.
If you are seeing problems, then take it out. Ensure all traffic goes to one server.
If you want HA for CAS, then you need a load balancer. Kemp or JetNexus are the most common ones I deploy.
ASKER
Hi, we use NLB,,,very simple load balancer. But I doubt that is the problem.
I'm sorry Simon but Round Robin is load balancing
All load balancers have a RR mode and even DNS round Robin provides Load Balancing. Maybe not HA in the case of 2010 but definitely load balancing in its most basic form
All load balancers have a RR mode and even DNS round Robin provides Load Balancing. Maybe not HA in the case of 2010 but definitely load balancing in its most basic form
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I am aware of all that but since users will be randomly distributed around the servers it is safe to assume that the load will be balanced even though it is a random distribution and not a smart one
The fact that it is not service aware makes it not suitable for high availability (which I said in my previous post)
The fact that it is not service aware makes it not suitable for high availability (which I said in my previous post)
Btw I'm not arguing with idea to bypass it or that it might be the source of the issue just debating the sentence that round robin is not load balancing
ASKER
OK....since we use an alias that points to the virtual node of the nlb cluster, is it acceptable to simply change the alias to point to one of the two physical nodes rather than the virtual one ? This way the change is very straight-through.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Should have enough info to answer the request