Appeal 2007-1595
Application 09/751,858
1 failures is the average of this time gap (Baum 451).
2 15. Time to repair is the time gap between the events of a machine failing
3 and its subsequent repair. Mean-time-to-repair is the average of this
4 time gap (Baum 451).
5 16. A time gap is a generalized form of a date gap (FF01).
6 17. Thus, Baum describes date gap analysis of events related to machine
7 uptime and time to repair as exemplified by a histogram chart to present
8 such analysis.
9 18. Baum describes applying a Kolmogorov-Smirnov (K-S) goodness of fit
10 test to the time gap data to test its distribution. This is a statistical
11 analysis based on the difference between the hypothesized distribution
12 and the actual data (Baum 452-453).
13 19. Baum applies the K-S test to machine uptimes (Baum, 452, Statistical
14 Analysis). Machine uptime is itself a measure of workload, because it is
15 the amount of time a machine works. The distribution function is
16 calculated for each individual data point in the sample (Baum, 453). The
17 number of data points equals the number of machines times the number
18 of data points per machine. The number of machines working is a
19 measure of the amount of work performed, and is also a measure of
20 workload. The K-S test then measures the greatest difference between
21 the distribution function and the data points in the sample. Like the
22 previous computation of the distribution function, this computation relies
23 on the number of machines and hence the workload. These differences
24 are themselves data that are modified during the computation, hence data
25 is being created and adjusted during the course of the computation based
12
Page: Previous 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Next
Last modified: September 9, 2013