timingstudiesforsamdeclaremetadatacommand.docVIP

Timing Studies for SAM declare Metadata command Tests to measure the capabilities of a v7_3_0 SAM dbserver were performed. These tests are documented in this note. This testing occurred on thefcdflnx3.fnal.gov. To perform these tests, a python script was run. The script used is in appendix A. These jobs used the SAM DB server user_int (v7_3_0) running on cdfsam05.fnal.gov. Table 1 lists the number of sections, type of test, start and stop times Table 1 List of SAM getMetadata tests performed Type of test # of job sections Submit time End time 1 Concurrent declare Metadata 10-50 jobs Tue Sep 6 00:18:00 2005 Tue Sep 6 02:04:00 2005 The purpose of these tests was to show the performance of the dbserver to concurrent declare metadata commands. The average and RMS for the declare metadata execution time as a function of the number concurrent commands. 9 cmds mean = 19.3256 RMS = 2.92697 10 cmds mean = 16.9128 RMS = 2.50036 19 cmds mean = 39.301 RMS = 5.75939 20 cmds mean = 39.3875 RMS = 5.64848 26 cmds mean = 76.0554 RMS = 7.78279 28 cmds mean = 65.671 RMS = 7.56844 29 cmds mean = 57.7923 RMS = 8.85945 30 cmds mean = 66.2936 RMS = 8.01189 46 cmds mean = 112.185 RMS = 11.7596 47 cmds mean = 115.753 RMS = 12.0578 48 cmds mean = 114.41 RMS = 10.752 49 cmds mean = 106.656 RMS = 10.2887 The next twelve plots show the distribution length of time each command took.  Figure 1 - Distribution of command execution time for declare meta data command as a function of the number of concurrent commands  Figure 2 - Distribution of command execution time for declare meta data command as a function of the number of concurrent commands  Figure 3 - Distribution of command execution time for declare meta data command as a function of the number of concurrent commands  Figure 4 Command execution time vs number of concurrent commands. The histograms in figures 1-3 are equivalent to slices of this plot. Figure 5 Profile histogram of the comman