### Table 14 Mean-time-to-failure (MTTF) and mean-time-to-recover (MTTR) are both measures that indicate reliability and availability. Generally, we define MTTF to represent the mean time over the total life of a system including periods of stable and less stable states. MTTR is the time a service is inaccessible to clients due to a failure. Neither MTTF nor MTTR reflects variance or distribution. We therefore propose that the estimated variance be included for both MTTF and MTTR. We denote the variances by V(MTTF) and V(MTTR), respectively. It also seems reasonable to state the maximum time of a repair; we denote it by max(MTTR). Statistical distributions for MTTF and MTTR can be defined, but we do not insist on them being a mandatory part of a contract.

1997

Cited by 10

### Table 4: Distribution of MTTR (days)

1995

"... In PAGE 6: ...navailable. Table 3 summarizes the results. Once again the distribution is strongly skewed, with the median time (just over eight hours) much shorter than the mean. Table4 . reports the percentiles, while Figure 5 shows the distribution and Figure 6 the cumulative distribution.... ..."

Cited by 45

### Table 4: Distribution of MTTR (days)

1995

"... In PAGE 6: ...Once again the distribution is strongly skewed, with the median time (just over eight hours) much shorter than the mean. Table4 . reports the percentiles, while Figure 5 shows the distribution and Figure 6 the cumulative distribution.... ..."

Cited by 45

### Table 4: Distribution of MTTR (days)

1995

"... In PAGE 6: ...navailable. Table 3 summarizes the results. Once again the distribution is strongly skewed, with the median time (just over eight hours) much shorter than the mean. Table4 . reports the percentiles, while Figure 5 shows the distribution and Figure 6 the cumulative distribution.... ..."

Cited by 45

### Table 4: Distribution of MTTR (days)

"... In PAGE 6: ...navailable. Table 3 summarizes the results. Once again the distribution is strongly skewed, with the median time (just over eight hours) much shorter than the mean. Table4 . reports the percentiles, while Figure 5 shows the distribution and Figure 6 the cumulative distribution.... ..."

### Table 8: Mean-Time-to-Repair for Speci c Sun 3 Models. Model MTTF Availability MTTR

### Table 7: Mean-Time-to-Repair for Speci c Sun 4 Models. Model MTTF Availability MTTR

### Table 2 A common formalization of availability is to define it as the relation between mean-time-to-failure and mean-time-to- repair. The assumption is basically that a system is either up or is being repaired and therefore the total elapsed time is the sum of MTTF and MTTR. Availability can then be defined as:

1997

Cited by 10

### Table 3: MTTF and MTTR of crashes of server types

2002

"... In PAGE 10: ... We set the MTTF and MTTR of soft and hard crashes of servers to reflect an extremely stress tested system. Table3 shows the values that we calibrated the crash module with. In addition, the crash module initializes a server once a week (i.... ..."

Cited by 20