Speaker
Description
Conclusions and Future Work
The portal offers several functionalities that the VOs appreciate. The foreseen next step is the development of a site flavour of the same portal that would provide the site administrators with a cross-VO view of their site's performance. In the longer run, SAM is looking into reusing parts of the portal for the visualisation of test results on the Regional Operation Center (ROC) level, when the new testing infrastructure is deployed.
URL for further information
http://dashb-sam.cern.ch
Keywords
Service Availability Monitoring (SAM), Site availability, Site reliability
Impact
In its current state, the portal already meets the requirements expressed by the VOs. It is available for the four major LHC experiments as well as the OPS VO. ATLAS, LHCb and CMS use the full panel of functionalities provided by the portal.
In order to present useful information to the VOs, an extension to the SAM database schema was needed. This extension holds information about the VO topology (tier repartition, VO-specific site naming convention). Only this part cannot be common to all the VOs and has to be updated regularly. For this purpose, a generic collector has been developed, but it is still up to the VOs to provide the information to be taken into account to maintain the topology information consistent.
Detailed analysis
The current version of the interface is composed of three main parts. The first part allows the VOs to view the SAM test results. The VOs are able to select the test results that are relevant for a given activity (e.g.: user analysis, data access). The second part allows VOs to define several groups of test types that they consider to be critical for a given activity or a given group of sites (e.g.: Tier1s only). The third part provides the VOs with metrics such as availability and reliability (as defined by GridView) based on the sets of tests previously defined.
This helps the VOs to identify and fix problems that can happen in a Grid infrastructure and to monitor the behaviour of sites.The portal is generic enough so that the exact same piece of code is executed for all the VOs. Therefore providing the interface for another VO is only a matter of configuration.
