Ibms Ras Philosophy - IBM p5 590 System Handbook

Problem Management Problem Management
Repair,
Concurrent Concurrent
Service
Repair Repair
Error Logging / Error Logging /
Diagnosis Diagnosis
Diagnose,
Reconfigure
Threshold / Threshold /
Prediction Prediction
Successful Successful
Recovery,
Fault Fault
retry
Masking Masking
Fault
avoidance
Figure 6-1 IBMs RAS philosophy
Both the p5-595 and p5-590 are designed to provide new levels of proven,
mainframe-inspired reliability, availability, and serviceability for mission-critical
applications. It comes equipped with multiple resources to help identify resolve
system problems rapidly. During ongoing operation, error checking and
correction (ECC) checks data for errors and can correct them in real time. First
Failure Data Capture (FFDC) capabilities log both the source and root cause of
problems to help prevent the recurrence of intermittent failures that diagnostics
cannot reproduce. Meanwhile, Dynamic Processor Deallocation and dynamic
deallocation of PCI bus slots help to reallocate resources when an impending
failure is detected so applications can continue to run unimpeded.
The p5-595 and p5-590 also include structural elements to help ensure
outstanding availability and serviceability. The 24-inch system frame includes
hot-swappable disk bays and blind-swap, hot-plug PCI-X slots that allow
administrators to repair, replace or install components without interrupting the
system. Redundant hot-plug power and cooling subsystems provide power and
cooling backup in case units fail, and they allow for easy replacement. In the
event of a complete power failure, early power off warning capabilities are
designed to perform an orderly shutdown. In addition, both primary and
redundant battery backup power subsystems are optionally available as well as
UPSs.
The p5-590 and p5-595 RAS design enhancements can be grouped into four
main areas:
Failure Failure
Recovery Recovery
Remote Remote
Support Support
Successful Successful
User User Redundant Redundant
Notification Notification
Policy Policy Failover Failover
Fault Isolation Fault Isolation
Failure Damage Containment Failure Damage Containment
Analysis Analysis
Unsuccessful Unsuccessful
Redundancy Redundancy
HW Retry HW Retry
Error Detection Error Detection
Base Hardware and Software Design Integrity Base Hardware and Software Design Integrity
Chapter 6. Reliability, availability, and serviceability
Failure Resilience Failure Resilience
Restart Restart
Recovery Recovery
Unsuccessful Unsuccessful
Software Software Damage Damage
Retry Retry
Control Control
Failure Data Capture Failure Data Capture
141