Monitoring the CMS Computing Infrastructure

Friederike Nowak
U Hamburg / CMS

At the LHC, each of the experiments will produce around a petabyte
of data per year. Managing and Processing these huge amounts of
data at one single computing center would be too cost and manpower
intensive. For this reason, the LHC experiments have decided to use
a computing model with geographically distributed resources. These
resources are provided by a worldwide grid, the WLCG (Worldwide LHC
Computing Grid). In the CMS computing model, the centers are
aranged in a hierarchic tier structure. The Tier 0, located at CERN,
is responible for the raw data from the detectors. The distributed
Tier 1's handle the reprocessing of the data, and on the Tier 2's,
the user's analyses will finally run. The reliability of these tiers
is crucial for all physics analyses, for the accessability of all
data will depend on the centers and the grid services. Therefore,
monitoring is an important part of running a site embedded in this
tiered layer.
This talk will give an overview of the CMS computing model. Because
DESY is hosting a Tier 2 center for CMS, it will also briefly cover
the monitoring of the computing centers, from the Tier 2 center's
point of view. As such a site in this Grid consists of many
components, monitoring tools are numerous and their surveillance is
highly time consuming. To get this under control, a meta monitoring
tool named 'Happy Face' was developed and is successfully operated at
four sites, one of it the DESY Tier 2.