We will move two applications to production (A and B); application A will have 9 thousands of Web users and B 4 thousands of Web users (total 13 thousands Web users). Each application is executing in separate cluster and each cluster with two LPARs. The applications are using DB2 JDBC driver type 2 and type 4 to access DB2 V8.1 on z/OS V1.8. We are moving to production these applications without do any previous Stress testing.

We have 4 Web Servers on z/OS (two per system V5.3). The applications are provided by third Part Company. The WebSphere Application Server for z/OS is V6.0.1 at Service level

Questions and if do you have any experience running J2EE applications on WebSphere Application Server for z/OS, please let me know, by example number of users, number of servants, threads, etc.

1. Can we support 13,000 Web users with 4 Web Servers on this sysplex environment?
2. We have the JVM HEAP size with 320 MB. Is this value right or not?
3. We have the settings for the minimum number of servants equals a 2 and a maximum of 6. Are those values the right ones?
4. What is the recommended value for workload profile (working threads): IOBOUND (in our case it assigns 9 threads) or LOGWAIT (40)?
The applications have EJBs and WAR components.
5. Which are the recommended values for number of connection for DB2 on Data Sources, for MQ, on Web Container?
6. To control the EC3 abends by timeouts, which value for the customer property "protocol_http_timeout_output_recovery" is recommended: SERVANT (default) or SESSION