LDAS CIT Database Insertion Rates
Test Period January 25-26, 2000
Input data Ilwd with dimensions 100
Platform metaserver.ligo.caltech.edu (SUN 450, 4 CPUs, RAM=1024MB, Solaris 2.7)
LDAS software ldas-0.0.8, IBM DB2 6.1 
Results are displayed in the order inserted.
Table #rows per block insert CPU time (secs) Wall Time (secs) Rate (#rows/sec)
process 100 0.39 0.653 153.14
process_params 100 0.25 0.309 323.62
frameset_chanlist 100 0.13 0.286 349.65
frameset_writer 100 0.13 0.469 213.22
frameset 100 0.24 0.317 315.46
frameset_loc 100 0.33 0.425 235.29
segment_definer 100 0.24 0.433 230.95
segment 100 0.14 0.193 518.13
summ_value 100 0.35 0.552 181.16
summ_statistics 100 0.31 0.456 219.30
summ_comment 100 0.43 0.579 172.71
summ_spectrum 100 0.41 1.148 87.11
filter 100 0.19 0.355 281.69
filter_params 100 0.26 0.330 303.03
gds_trigger 100 0.37 0.580 172.41
sngl_inspiral 100 0.24 0.409 244.50
sngl_burst 100 0.21 0.401 249.37
sngl_ringdown 100 0.25 0.447 223.71
sngl_unmodeled 100 0.25 0.447 223.71
sngl_unmodeled_v 100 0.17 0.259 386.10
sngl_dperiodic 100 0.26 0.457 218.82
sngl_datasource 100 0.25 0.346 289.01
sngl_transdata 100 0.44 1.130 88.50
multi_inspiral 100 0.25 0.459 217.86
multi_burst 100 0.21 0.430 232.56
coinc_sngl 100 0.50 0.853 117.23

Summary:

  1. Insertion rate is much improved due to faster machines, direct link to DB2 server instead of via openlink software and improved code efficiency.

  2. Insertion for tables summ_spectrum and sngl_transdata are slower because these tables contain blob data.