2. USAGE CONSIDERATIONS › 2.3 Use Within the Information Structure › 2.3.4 Installation Environment Analysis › 2.3.4.1 Environment: Reliability › 2.3.4.1.4 Reliability Analysis
2.3.4.1.4 Reliability Analysis
Mean-Time-Between-Failures (MTBF) is one of the most widely
used measures of device reliability. MTBF figures are used
to schedule routine maintenance, determine the requirements
for backup equipment, compare the reliability of a given
device to industry or installation standards, or aid in the
selection of competing equipment models.
Tracking the rate of change of MTBF statistics over time can
also yield valuable results. Decreasing MTBF values can
foreshadow the end of the practical useful life of a device.
Conversely, increasing MTBF values are sometimes encountered
in installations which utilize heavily microcoded,
"intelligent" devices such as point-of-sale or financial
controllers. These devices may experience random hard
failures due to undiscovered microcode errors. A significant
increase in MTBF following a microcode update may indicate
that such a "bug" was fixed.
MTBF statistics are easily reported using the data retained
by CA MICS in the NPDA Event (SNTNEF) File. Each entry in
this file in the DETAIL time-span records the occurrence of a
permanent error on a specific device. In the higher
timespans (DAYS, WEEKS, MONTHS, and YEARS), the number of
failures of a given type are summarized for each device in
the variable Number of Events (NEFNREVT). Since basic MTBF
statistics are based on the total number of failures, the
data must be summarized across all types of failures. This
can easily be done using the CA MICS Summarization Facility
as follows:
%INCLUDE INCLLIB(#SNTSUM);
MACRO _BY NEFFRNAM %
MACRO _BREAK NEFFRNAM %
PROC SORT DATA=_dSNTM.SNTNEF01 OUT=MTBF; BY _BY;
RUN;
DATA MTBF; SET MTBF; BY _BY;
_NEFSUM
RUN;
DATA MTBF (KEEP=NEFFRNAM MTBF);
FORMAT MTBF TIME.;
LABEL MTBF=MEAN-TIME-BETWEEN-FAILURES;
MTBF = (ENDTS-STARTTS) / NEFNREVT;
RUN;
PROC PRINT LABEL;
VAR NEFFRNAM MTBF;
TITLE1 'MEAN-TIME-BETWEEN-FAILURE STATISTICS';
RUN;