SC4 / pilot WLCG Service Workshop

This talk will try to identify the key issues that we need to understand as a result of this workshop, including:

- Discussing and obtaining draft agreement of the detailed Service Plan for 2006 (what services are deployed where, schedule, agreed functionality that is delivered by each release etc.)

- Data flows / rates (reprocessing at Tier1s, calibration etc. (e.g. the TierX-TierY ones...)

- Corresponding impact on Tier1s and Tier2s

The Tier0-Tier1 rates, as well as first pass processing, are assumed to be well known ...

The status and timeline of the LCG Optical Private Network (OPN) is presented.

This session will cover recent and foreseen enhancements to the FTS s/w and service. (20')

It will also provide an opportunity for sites to express their requirements in terms of operation, manageability and so forth.

This session will summarise recent and foreseen enhancements to the DPM & LFC software and services (20')

It will also cover the site perspective from a deployment / service point of view.

Deployment schedule of CASTOR2 and CASTOR SRM across the sites, including:

- proposed h/w configuration to be used at CERN for SC4;

- specific CASTOR and SRM 2.1 features provided by each release, 

- testing schedule, site & experiment validation etc.

- Summary of dCache workshop.

- Deployment schedule of dCache 2.1 SRM across the sites, including specific features provided by each release, testing schedule, experiment validation etc.

By definition, the various components cited above must interoperate. We should identify any outstanding issues and produce a plan to resolve them.

List of requirements for setting up a production service - the "service dashboard" / "checklist"

By the time of this workshop, all main Grid services at CERN will have been redeployed to address the MoU targets.

This session covers the various techniques deployed as well as initial experience with running the services.

Recommendations for sites deploying the same services will be offered.

Experience with monitoring services using the Site Functional Tests

A proposal for how component level testing could be aggregated into MoU-level services; metrics; reporting etc.

In this session we attempt to reach draft agreement on the detailed Service Schedule for 2006.

This includes:

- The list of services deployed at which sites;

- The service roll-out schedule;

- The service roll-out procedure (which must include both pre-testing / certification by the deployment teams as well as prolonged testing in a production environment (the Pre-Production System assumed) by the experiments.

- The operations model for 2006 and roll-out schedule (from 'SC operations' to standard operations - what, if anything, remains particular to WLCG production?)

- Coordinating the various mailing lists, announcement mechanisms, out-of-hours coverage...

- The support model for 2006: moving from stop-gap support lists setup for SC3 to GGUS support; time-table, scope etc.

A discussion on the target service levels. The baseline is given in the attached planning document.

All sites - at least the Tier1s and DESY - should come prepared to report on service levels and availability at their site.

Conclusions of the workshop held at CERN on February 6th:

Agenda :
http://agenda.cern.ch/fullAgenda.php?ida=a058495

The workshop goals are to review the status of the database project milestones on the tier 1/0 side and on the experiment side.

This boils down to

- to what extent are the production database servers and frontier installations at T0 and T1     becoming available? What problems setup have been encountered/ solved? Which steps still need to be completed before opening the first distributed LCG database setup for users from the experiments in March? How can the remaining Tier 1 sites join/participate during the first production phase (March-Sept) in order to be ready to also provide database services for the full deployment milestone in autumn?

- to what extent has the development of the main applications (conditions data and other apps to be tested as part of SC4) been completed? This includes the definition of concrete detector data models including refined estimates for volume and access patterns, connection between online and offline, connection between T0 and T1 (reports from the various distribution tests).

Service and deployment issues...

Target sites, service levels, monitoring, procedures, etc.