Hi All,
We are having problems with the deplopyment of a multi-host configuration of BizTalk 2004 instances. In fact, in this configuration, the system reacts really slow. Details are below. Can you help?
When migrating our solution to an environment with multiple BizTalk application servers and a clustered SQL database, we are seeing very long service instance execution times for sending to an MSMQT send port. We have five BizTalk application servers at this moment and two of these servers run the host containing the MSMQT adapter, and there is a virtual IP address for the hardware load balancer pointing to these two servers.
The service instance of the send port in HAT indicates execution times of 30 seconds to 40 minutes (!!). The send port uses the standard Microsoft XmlTransmit pipeline.
We referred to page 453 of Scott Woodgate's BizTalk book that indicates queue names should not use the OS:.\ syntax but rather should use TCP:x.x.x.x\PRIVATE$\ (where x.x.x.x is the IP of the load balancer), and we made this change, but this seems to have made the problem worse.
We still experience problems if we point the load balancer to just one server and ensure the relevant hosts are only running on that one server.
Have we missed something about MSMQT configuration in such a multi-server environment? Can we do it with a hardware load balancer or do we have to use NLB? Are there other things we have missed?
Does anyone know what might be causing this?
We are having problems with the deplopyment of a multi-host configuration of BizTalk 2004 instances. In fact, in this configuration, the system reacts really slow. Details are below. Can you help?
When migrating our solution to an environment with multiple BizTalk application servers and a clustered SQL database, we are seeing very long service instance execution times for sending to an MSMQT send port. We have five BizTalk application servers at this moment and two of these servers run the host containing the MSMQT adapter, and there is a virtual IP address for the hardware load balancer pointing to these two servers.
The service instance of the send port in HAT indicates execution times of 30 seconds to 40 minutes (!!). The send port uses the standard Microsoft XmlTransmit pipeline.
We referred to page 453 of Scott Woodgate's BizTalk book that indicates queue names should not use the OS:.\ syntax but rather should use TCP:x.x.x.x\PRIVATE$\ (where x.x.x.x is the IP of the load balancer), and we made this change, but this seems to have made the problem worse.
We still experience problems if we point the load balancer to just one server and ensure the relevant hosts are only running on that one server.
Have we missed something about MSMQT configuration in such a multi-server environment? Can we do it with a hardware load balancer or do we have to use NLB? Are there other things we have missed?
Does anyone know what might be causing this?