Facilities

Andrew Melo
Vanderbilt University

Facilities Co-Leader

David Mason
Fermilab

Facilities Co-Leader

Overview

Facilities are responsible for maintaining the necessary computing services and software stack for proper functioning of Fermilab as the U.S. CMS Tier-1 computing facility to CMS. This includes on-call response personnel and capability for storage, network, batch and other services to maintain the required level of service for the WLCG. This facility fulfills the resource pledge made by U.S. CMS to the WLCG to help meet CMS computing resource requests.

Facilities leads the U.S. CMS security effort and works to further improve the security procedures of the U.S. CMS sites and to maintain coordination with domestic and international computing security efforts.

The facilities are members of the Open Science Grid (OSG) consortium.

Services

distributed batch processing
archival disk and tape storage
data transfer
monitoring
CMS software and services.

CMS Tier-1 Facility

As a CMS Tier-1 facility, the site supports:

event processing and reprocessing
event simulation production
user analysis
data hosting
archival storage
local data access
streaming data access over WAN

Tier 2 support

There are 7 Tier-2 sites hosted at U.S. CMS institutes: Caltech, University of Florida Gainesville, MIT, University of Nebraska Lincoln, Purdue University, University of California San Diego, and the University of Wisconsin Madison. The Tier-2 sites provide computing and disk storage resources to enable physics analysis research and to fulfil our committments to the CMS experiment.

As CMS Tier-2 facilities, the sites support many of the same functionalities as the Tier-1 with the exception of archival storage and with a much stronger focus on physics data analysis by researchers. The support model is business-hours only, but often site administrators will put significant effort after hours and on weekends to keep our Tier-2 sites as some of the best-functioning in the CMS experiment.

Tier 2 Usage Chart — Total (green) and central production (red) processing usage at the 7 U.S. CMS Tier-2 sites monthly since 2016 in core-months, compared to the total purchased compute capacity, in batch slots (black). The difference between the total and central production usage is physics analysis. Note that before August 2017, production usage fluctuated due to inconsistent queue depth, while since then the sites have been consistently and fully utilized. Some months total usage is more than the purchased processing capacity due to opportunistic cycles used at campus clusters or HPC’s.

Tier 3 support

Tier-3 sites are typically smaller computing and/or storage facilities hosted at universities, with varying degrees of connectivity to the overall CMS infrastructure and range from small storage clusters for physics analysis to fully-functional Grid sites. We have a support team in U.S. CMS which helps the site administrators at Tier-3’s, who may not have extensive knowledge of CMS software or practices, to integrate their sites into the larger CMS CI.

Networking and Security

The facilities area also oversees operations in U.S. CMS in the fields of security and networking. For security we coordinate vulnerability announcements with the OSG for our sites and provide documentation for best practices. For networking we participate in various network capacity and capability studies in planning future needs.