Bookmark and Share

Recent Posts

sig·nif·i·cance – System Management Facility

November 14, 2016

This week, I am continuing this series on significance (the quality of being worthy of attention or importance). For this post, I want to focus on a component of z/OS called SMF. As a new application developer, I had no exposure to SMF. It was something developers didn’t know about because we were focused on other things like using COBOL and VSAM to code a solution to the specs given to us by our analyst. Later, when I became a system programmer trainee, I became aware of the enormous significance of SMF.

Implementation of a Big Idea
SMF is the realization of a big idea. SMF provides a repository with support routines that collect data about important activities that are taking place on the mainframe (e.g., job initiation, step termination, job termination, input and output data set activity, and many other activities). This important data is captured in over 100 different types of SMF records.  SMF gathers baseline information and many other products contribute to the collection make the data much more complete and useful.

What Products?
The RMF uses SMF to add performance records to its repository. The data in these records focuses on processor, paging, channel path, page and trace data set activity, virtual storage and I/O queuing including data on serialization. That isn’t the whole story because RMF has real-time monitoring that can include data from distributed systems (RMF XP).

RACF collects data for type 80 records (Security Produce Processing). These records are written to record security issues (e.g., password violations or denied resource access attempts and other problems). ISV products like CA TopSecret and CA ACF2 provide similar information to SMF.

IBM DB2 writes DB2 statistics, accounting and performance records. For IMS, IBM Guardium S-TAP for IMS reports events that are being collected by SMF from more than a dozen different record types. CICS Transaction Server writes transaction data collected at event monitoring points into the SMF data repository. There’s no specific transaction-level data created for IMS DC (now called IMS TM) in IBM SMF records however RMF can supply resource data for the IMS workload at the address space, service classes and reporting classes level.

IBM WebSphere MQ collects both statistics and accounting data. IBM WebSphere Application Server for z/OS writes the type 120 record that contains performance statistics 

Where Does the Data Go?
Since many products generate data, it’s interesting to note how the data is recorded. SMF can record data in two ways. One technique buffers the SMF address space together with a set of VSAM data sets  to use when a buffer fills up. A different method uses log streams. SMF utilizes the system logger to record collected data, which improves the writing rate and avoids buffer shortages. In both cases, the data is used for reporting where companies collect data for daily, weekly and monthly report processing.
It Didn’t Happen All At Once
Today’s implementation of SMF didn’t happen all at once but rather evolved and grew over the decades. However, the early designers of MVS realized the importance of gathering this data and were not daunted by the challenge of collecting this data to meet the needs of planners and problem solvers. 

Posted November 14, 2016 | Permalink

comments powered by Disqus