GRIDKA BABAR Statistics

CPU and user statistics 2004, 2005, 2006 and 2007, 2008, 2009, 2010, 2011, 2012, accumulated data (since 2002),
import/export (past years), SP6 MC production, SP8 MC production, SP9 MC production, SP10 MC production, CPUS and disks, Ganglia.

CPU Usage 2012

Fairshare Statistics


abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: number of different analysis users per week, starting 2004.
Click the picture for a small version without letter scaling.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage 2011

abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage 2010

abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage 2009


abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage 2008


abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage 2007

abscissa: date (start of corresponding week)
ordinate: CPU time spent per week. The total is subdivided into a fraction for users (analysis), for MC production, for Skim production, and for the rest (including at present mostly Admin jobs).


abscissa: date (start of week)
ordinate: CPU time spent by analysis users per week. The total is subdivided into a fraction for users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: wall time spent by different users per week. The total is subdivided into a fraction from analysis users and from MC production.


abscissa: date (start of week)
ordinate: number of different analysis users per week. The total is subdivided into a fraction of users from German groups and from other BABAR groups.


abscissa: date (start of week)
ordinate: Number of jobs submitted within one week. The total is subdivided into a fraction for users from German groups, users from other BABAR groups, and MC production jobs.


abscissa: date (start of week)
ordinate: weekly maximum and average virtual memory of analysis user jobs, and average virtual memory of Monte Carlo production jobs.


abscissa: date (start of week)
ordinate: weekly average queue waiting times of analysis user jobs, and of Monte Carlo production jobs.

all diagrams shown above are created from this data table

CPU Usage since 2002


abscissa: date (start of week)
ordinate: weekly CPU hours since start of 2004, broken down to different user classes, and scaled to the BABAR CPU performance unit.




diagrams shown above are created from data tables before 2005, and for 2005, 2006, 2007, 2008, 2009.

Gridka Data Import/Export

abscissa: date
ordinate: GB imported on this day


abscissa: date
ordinate: GB exported on this day, broken down to MC and Skim data


abscissa: date, with 7-day-grid
ordinate: data transfer rate (kB/s), given as ratio data / transfer time in (bluish) green and as ratio data / 24hours*3600s in brown or orange

created from this data table

SP8 Monte Carlo Production in Karlsruhe, Dresden and Dortmund

abscissa: 1st day of week in question
ordinate: million events per week, subdivided into fractions produced at Gridka, Dresden, Dortmund and elsewhere


abscissa: date
ordinate: million events produced up to date, subdivided into fractions produced at Gridka, Dresden, Dortmund and elsewhere


abscissa: 1st day of week in question
ordinate: fraction from total MC production produced at Gridka, Dresden and Dortmund


plots created from BABAR table

SP9 Monte Carlo Production in Karlsruhe, Dresden and Dortmund

abscissa: 1st day of week in question
ordinate: million events per week, subdivided into fractions produced at Gridka, Dresden and elsewhere


abscissa: date
ordinate: million events produced up to date, subdivided into fractions produced at Gridka, Dresden and elsewhere


abscissa: 1st day of week in question
ordinate: fraction from total MC production produced at Gridka and Dresden


plots created from BABAR table

SP10 Monte Carlo Production in Karlsruhe, Dresden and Dortmund

abscissa: 1st day of week in question
ordinate: million events per week, subdivided into fractions produced at Gridka, Dresden and elsewhere


abscissa: date
ordinate: million events produced up to date, subdivided into fractions produced at Gridka, Dresden and elsewhere


abscissa: 1st day of week in question
ordinate: fraction from total MC production produced at Gridka and Dresden


plots created from BABAR table

Hardware

Date PCs CPUs RAID Storageour share
Oct 2004 535 1070 >270 TB netto50 TB
Apr 2005 642 1284 >270 TB netto50 TB
May 2005 642 1284 70 TB
Jul 2006 78 TB
Sep 2006 87 TB
Oct 2006 96 TB
Nov 2006 91 TB
Jan 2007 96 TB
Mar 2007 1000 2700 1790 TB 109 TB
Oct 2007 117 TB
Apr 2008 151 TB
May 2008 146 TB
Jun 2008 125 TB
Nov 2008 179 TB
Jul 2009 >185 TB
Nov 2010 >200 TB

Disk space is in binary units, i.e. powers of 1024. Details are here.

BABAR disk space list

Roland Waldi, last update 24 Jan 2012