SEECS vs SLAC
The pingtable.pl results measured from SLAC and from SEECS differs in some cases. For example compare:
- http://pinger.seecs.edu.pk/cgi-bin/pingtable.pl?file=average_rtt&by=by-node&size=100&tick=monthly&from=PK.AUP.EDU.N2&to=PK.UPESH.EDU.N2&ex=none&dataset=hep&percentage=any with
- http://www-wanmon.slac.stanford.edu/cgi-wrap/pingtable.pl?file=average_rtt&by=by-node&size=100&tick=monthly&from=PK.AUP.EDU.N2&to=PK.UPESH.EDU.N2&ex=none&only=all&dataset=hep&percentage=any.
For October 2010 SEECS reports 0.777 while SLAC reports 0.985. Looking at the October daily results:
- http://pinger.seecs.edu.pk/cgi-bin/pingtable.pl?file=average_rtt&by=by-node&size=100&tick=daily&year=2010&month=10&from=PK.AUP.EDU.N2&to=PK.UPESH.EDU.N2&ex=none&dataset=hep&percentage=any
- http://www-wanmon.slac.stanford.edu/cgi-wrap/pingtable.pl?file=average_rtt&by=by-node&size=100&tick=daily&year=2010&month=10&from=PK.AUP.EDU.N2&to=PK.UPESH.EDU.N2&ex=none&only=all&dataset=hep&percentage=any
It is seen that SLAC is missing data from October 2nd 2010. Otherwise the values are identical. Since SLAC and SEECS gather the data from the monitoring sites independently. There will occasionally be differences since the remote node may be inaccessible when one of the monitoring hosts is gathering the data. If one is pedantic one can use getdata.pl to re-gather the missing data. For example in this particular case to re-gather the missing data from SLAC one would use:
getdata.pl pinger.aup.edu.pk 2010-10-02
Table vs pingtable.pl
Observation
Sadia was working on report and this is what she came across. She was using pingtable.pl for data of 12 months. She found that values are contradictory between table.pl and pingtable.pl. Compare:
- http://pinger.seecs.edu.pk/cgi-bin/table.pl?from=Pakistan&to=Pakistan&file=throughput&date=2010-11 with
- http://www-wanmon.slac.stanford.edu/cgi-wrap/table.pl?file=throughput&by=by-node&size=100&tick=monthly&from=Pakistan&to=Pakistan&ex=none&only=all&dataset=hep&percentage=any
They purport to show same statistics. For example from COMSATS to BUITMS throughput is 7829.42 for the month of November. However pingtable.pl is showing 7457.728 whereas table.pl is showing 5523.796.
Reason
The way the two scripts calculate their stats is different. Table.pl averages the average values given for the days of the month. Pingtable.pl gets its data by going back to the average hourly values and averaging them.
The reason for the difference is that the data pingtable.pl uses is pre-calculated, thus the number of data points that have to be crunched will not impact the interactive execution time of the pingtable.pl CGI script. Table.pl on the other hand does not use the precalculated data instead it does the calculation on the fly. To speed it up rather than taking every hours data (i.e. 24 times as much data) it starts from the daily average. There is no reason someone with energy etc. could not modify table.pl to use the precalculated monthly data as pingtable.pl does, so the results would be identical (and table.pl should execute more quickly).