Detecting and Diagnosing Problems When z/OS Thinks It Is OK (includes PFA User Experience)

Project and Program: MVS, MVS Core Technologies

Tags: Proceedings, 2012, SHARE in Atlanta 2012

The presenter will discuss the multiple capabilities which are available on z/OS to detect and diagnose soft failures

• Describe soft failure detection
  
  
  
  • Built into z/OS component like XCF stalled member detection
  
  
  • Provided by health checks
  
  
  • Provided by z/OS PFA
  
  
  • Provided by other vendor products
  
  
  
  


• Highlight the kind of problems each different type of soft failure detection is good at and not good at
  
  
  
  • Machine time scale vs human time scale
  
  
  • Location in the stack
  
  
  • Detectable by performance metrics vs non performance metrics
  
  
  
  


• Insight from building PFA to help reduce impact of soft failures
  
  
  
  • Automation of alerts is key
  
  
  • z/OS can survive / recover from most soft failures
  
  
  • Most metrics are very time sensitive
  
  
  
  

Robert Abrams, IBM Corporation; Sam Knutson, GEICO

Back to Proceedings File Library