Inside Products has presented these papers or presentations at conferences including SHARE or Computer Measurement Group (CMG). Please select the link for the paper to see the abstract. To download the papers of your choice, please sign our guestbook at Guestbook. You may choose to download the presentation you wish.

To see articles published in zJournal or Technical Support, please go to: Articles.

• Ten Commandments of TCP Performance
• TCP/IP Performance Analysis for Dummies
• TCP/IP Network Health Check
• Looking Inside TCP, FTP and Socket Applications
• TCP Detective Stories
• Performance Monitoring with the New zOS SNMP MIB
• Chargeback for Mainframe TCP/IP Networks

The Ten Commandments of TCP/IP Performance are a distillation of hard-won experience. Monitoring and tuning TCP networks on the mainframe is complex for the basic reason that each network is a mixture of many applications and pieces of hardware.

What are the basic things you need to know to do a analyze performance for your mainframe TCP/IP system? We will discuss the well-known TCP/IP applications (FTP, Telnet, SMTP) and the basic commands needed to diagnose problems on the stack and network.

This presentation will also cover the crucial parameters to check for the health of the TCP/IP stack, how to check them, and what they mean. A guide to tasks which comprise Level 1, 2 and 3 TCP/IP diagnostics will be presented.

TCP/IP Network Health Check

Is your network tuned up? Are your TCP Profile parameters defined correctly? What errors is your TCP/IP stack encountering? Have you looked at the CPU time or queuing for your socket applications?

In this presentation, we describe how you may be able to do a step-by-step performance check of the TCP/IP network. We will do the following:

Looking Inside TCP, FTP and Socket Applications

Learn how you can look inside your TCP system tasks and socket applications and OMVS processes. Do you wonder about what percent of CPU your TCP/IP and socket attached applications are taking and if there are places that can be changed both in the system and in applications to save CPU time? Would you also like to measure paging, SRM and queues?

If, as many others, you are concerned with saving CPU cycles, this session will suggest a methodology to:

• Benchmark your applications
• Make tuning / performance changes
• Measure again to see if CPU time is saved either absolutely or relative to other applications.

A special section on FTP will be presented showing and grouping the MVS jobs and USS processes which are created when an FTP is done.

TCP Detective Stories

What kind of problems have you had in your TCP/IP network? Are you wondering what problems other people are really running into in TCP/IP implementations? We will discuss a number of real problems caused by the complex interaction of application software, network hardware, and TCP/IP parameter settings and how people have solved them.

The second part of this session will cover some real-life monitoring scenarios and the techniques various companies are using to manage critical business applications running over their TCP/IP networks.

Performance Monitoring with the New zOS SNMP MIB

The new z/OS TCP/IP stack (z/OS 1.2 – z/OS 1.5) has a greatly enhanced SNMP MIB to allow the systems programmer to do diagnostics for the TCP/IP stack and sessions. The SNMP MIB includes hundreds of variables to analyze the core IP protocols (IP, TCP, UDP, and ICMP) plus many other variables not available by any other means. Data such as backlog for Listener sessions or IP packets dropped for CSM memory shortages provide invaluable sources of data to monitor for z/OS performance and troubleshooting.

The z/OS SNMP MIB also contains information to analyze OSA Express performance and throughput. Some information is not available through Netstat commands or via SMF records. OSA Express provides high speed access for FTP and other critical traffic - knowing the throughput and utilization of this expensive and critical resource is mandatory for a systems programmer.

The many variables and their uses will be discussed in this session.

Chargeback for Mainframe TCP/IP Networks

Many companies are curious about billing for TCP/IP usage on the z/OS platform to users or applications for their usage of associated resources. TCP/IP started tasks and the services which rely on TCP/IP (such as FTP, Tivoli Service Manager, MQSeries, Websphere) use scarce and expensive host CPU time.

Often applications are coded with seemingly little regard for the amount of CPU usage, network traffic or other scarce resources. TCP/IP resource usage seems to be growing exponentially.

We will discuss topics such as: what are the factors which may be monitored and billed? Should peak time usage and other methods be used to help in conservation of critical resources? What is "fair?" Charge back should be for the most scarce resources.