Living With Enterprise Extender: EE Problem Determination in a z/OS Environment

by Gordon Webber
November 1, 2006

• Session partners
• IP addresses and ports used
• Routes
• Link and connection states
• Congestion/flow control
• Fragmentation and retries
• Session states
• Error codes
• SNA control commands
• SNA data.

In the Network Header (NHDR) portion of the NLP, there are two slowdown indicators. Initially, these are clear, but if any intermediate node along the route becomes congested, it can set one of the slowdown indicators. When this NLP arrives at the destination RTP endpoint, the receiving host will adjust the data sending rate on the RTP connection. This is a function of ARB flow control. Investigation of these flags is useful when slow network performance is suspected. This may be a transient condition, highlighting, again, the need to trace these events.

Other potential problem areas include:

• Broadcasts may occur during APP N Locates (especially for LEN devices where no CP-CP sessions can be established— therefore, especially in migration scenarios; i.e., subarea-to-APP N).
• Things such as RUsize, BTU sizes, MTU sizes (IP network), MLS sizes (APP N networks), COS selection, and pacing window sizes all affect the performance; most of these can be observed by tracing.

Tracing will reveal information such as:

• LOGMODES: RUsizes
• BROADCASTS: observe the packets
• FLOW CONTROL: shows how well ARB is working and if the original flow rate was appropriate.
• FRAGMENTATION: will be highlighted by the trace, and indicates bad packet to MTU ratio, or maybe an error in a network device (router) configuration.

Summary

EE is a great solution for allowing mainframe SNA traffic to be carried across IP networks. This affords an organization maximum advantage by allowing legacy applications to continue to be used (avoiding costly rewrites) while benefiting from advances in IP network technology.

EE can be implemented in many situations from relatively simple scenarios to complex heterogeneous networks. The ability to deliver and maintain high levels of service is paramount. Equally important, in the event of a problem, is the ability to rapidly pinpoint the root cause. This requires tools and staff with the appropriate skillsets to enable effective network investigation. A great deal of work and knowledge is required to correctly extract and interpret this information.

EE makes unique demands on the network professional. It requires a level of knowledge that encompasses multiple disciplines across both TCP/IP and SNA networking protocols. In the graying workforce, SNA skills are dwindling, which makes it all the more important to find the solutions that act as expert systems with built-in SNA and IP knowledge. Z