Clustering vs NonClustering with SAP R3 Adapter

Are other companies running IS 4.6 servers clustered and
are utilizing the SAP R/3 Adapter?

Curious, because SAP Adapter manual definitely does not
recommend clustering. We initially had our environment
installed that way and ran into strange (unexplained)
troubles. We are now beginning testing having two
unclustered IS servers with separate SAP R/3 adapter
installed on each.

Another question, can you have the same SAP listener defined
and listening on more than one IS server? Will this confuse
SAP, or will it create a pseudo load-balancing effect?

Thanks for any experience/knowledge!!

We are currently in the proof-of-concept stage with this configuration. We have successfully tested the use of SAP R/3 with two clustered Integration Servers.

The key is to use SAP load balancing. To do this, the adapters on both integration servers must be identical, including aliases and listeners. No other configuration on webMethods is necessary.

In SAP, you must set up two ports. Each port uses the same progId, but points to a different server. Our SAP guy said this was trivial to set up.

Once you have this configuration down, SAP will automatically load balance/fail-over between the two ports. If both IS’s are down, IDOCs will queue in SAP until one of them comes back up.

Hope this helps…

J Silver,

This is very interesting and I was under the impression that it was not possible. Do you think that you could elaborate?

Specifically, what do you mean by “you must set up two ports”? How does the “port” in SAP point to a different IS when both have the same program ID?

Our weakest link in our fail-over/load balanced environment is currently this issue of having one RFC Destination/Program ID listener running on two identical machines.

Would appreciate any feedback!

Thanks,

ajandja

This post is for reference. I responded to Andrew quite a while ago but thought this info would be useful to others as well… I’m not an SAP guy, so if there are some SAP instructions that don’t make sense its probably because I’m unwittingly using terminology specific to our implementation.

In SAP:

  1. Configure a port for each Webmethods server. For example:

Port - Description - Logical Destination
A00001 - DEV_100 webMeth 112 - DEV_WEBMETHODS_LAB_112
A00002 - DEV_100 webMeth 113 - DEV_WEBMETHODS_LAB_113

  1. Configure an RFC destination - type TCP/IP - using the same program id for each logical system added. The Port logical destination must match the RFC destination and the program id is the same for each RFC destination. For example:

Entry 1:
RFC destination - DEV_WEBMETHODS_LAB_112
Connection type - T TCP/IP connection
Registration Pgm ID - DEV100WMSERV112
Description - Webmethods server 112

Entry 2:
RFC destination - DEV_WEBMETHODS_LAB_113
Connection type - T TCP/IP connection
Registration Pgm ID - DEV100WMSERV113
Description - Webmethods server 113

WebMethods Configuration:
RFC destinations in the SAP alias on both servers will be identical.