wiki:Obsolete/Web/CaseStudies

Some Sites using Quattor

This page describes existing quattor deployments. For recent status reports from sites, please refer to the minutes of the quattor workshops.

Computer centres

quattor is being successfully used to manage large single-site installations comprising thousands of machines running a huge range of services. The flexibility of the quattor architecture allows standard scalability techniques (e.g. use of proxy servers) to be used for such large deployments.

CERN

CERN Computer Center The CERN computer centre currently manages over 7000 machines providing the computational and storage power needed to handle the vast amount of data due to flow from experiments on the Large Hadron Collider (wikipedia page). An overview of the CERN deployment can be found in the 2005 article ELFms at CERN.

CNAF

CNAF Storage Image CNAF is the National Center of INFN (Italian National Institute for Nuclear Physics) for Research and Development activities in Information Technologies applied to High Energy physics. Since 2005, CNAF has been the Italian Tier-1 center for LHC experiments and a reference computing center for several other experiments. As of November 2008, it provides distributed access to about 7MSI2K of CPU power, 2.3 PB of disk and 3PB of tape space. CNAF system administrators uniformly manage more than 1000 nodes (most of them running the EGEE middleware) with Quattor.

UAM

The high-energy physics group at Universidad Autónoma de Madrid (UAM) hosts a Tier-2 site for the LHC Computing Grid. Besides that, it runs a separate grid infrastructure at nights, using the spare computing power on the University's seminar rooms. The operating systems used for teaching purposes (Windows, SuSE Linux) are not managed by Quattor but by Rembo. Instead a partition is granted by UAM IT department.

Quattor installations can co-exist with other operating systems, and thus we get ~300 machines running 24/7, and over 1000 running during non-working hours.

Nikhef

Nikhef is the National Institute for Subatomic Physics in Amsterdam (Netherlands). Nikhef participates in various grid projects, such as EGEE and BigGrid, and is one of the computing centers in the Dutch Tier-1 for LHC experiments. The grid infrastructure at Nikhef, which consists of 300 hosts, is managed using Quattor.

STFC Rutherford Appleton Laboratory

The group responsible for running the UK's Tier-1 site at the STFC Rutherford Appleton Laboratory as part of the GridPP Project have been using Quattor to deploy and manage all new and redeployed hosts since late 2009.

As of January 2011 they are managing over 1200 hosts with Quattor.

Regional grid infrastructures

Large grid infrastructures such as EGEE are increasingly organised along regional lines. Independent sites (universities, research labs, etc.) collaborate to develop shared configuration parameters and services, spreading the administrative burden across multiple sites.

Grid-Ireland

Grid-Ireland Map Grid-Ireland is the national computational grid for Ireland, and is made up of eighteen sites managed from the Operations Centre based at Trinity College Dublin. Over 400 machines are now managed using quattor, including a 768-core cluster at TCD used for processing jobs from EGEE and Grid-Ireland users.

Grid-Ireland makes extensive use of Xen virtual machines to host services. All of these VMs are automatically configured using quattor.

BEgrid

GRIF

GRIF is a site of EGEE grid in Paris region (France). It is made up of six subsites in different geographical locations but is only one site in the GRID. It is particular one of the largest T2 of WLCG, the computing infrastructure for CERN LHC. More than 600 machines, including compute servers, storage servers with a total amount of more than 1 PB, non-grid servers and desktops, are managed by a team of 20 people located at the different sites. Quattor is critial site at GRIF to ensure both consistency of the whole site despite its distribution and allow every body to participate to the management of the global site, according to its expertise, leveraging the manpower available and avoiding duplication of efforts.

HellasGrid

HellasGrid is the national computational grid for Greece. Greek sites are currently using Quattor mainly for the base node installation (bootstrap). Sites hosted by GridAUTH are totally managed by Quattor. Till now GridAUTH's quattor server manages 2 Tier-2 sites with a total number of 150 nodes. The current Quattor managed infrastructure include both region core services (i.e. WMS, VOMS, Hydra servers, Web servers) and site services (i.e. UI, WN, CE, SE, MON).

Industry

Banking

Since November 2008, a financial institution in North America is using Quattor to manage a datacenter of 6,500 grid nodes. Quattor manages not only the compute nodes, but also many of the core infrastructure services, with a view to gradually increase the footprint of Quattor management into more of the core.

Philips Research

Philips Research participates in various grid projects, one of which is BiGGrid. A part of BiGGrid is hosted by Philips Research at the High Tech Campus (HTC), Eindhoven, The Netherlands. The main grid infrastructure of about 1600 cores at HTC Eindhoven is managed through Quattor.

Last modified 13 years ago Last modified on Apr 21, 2011, 3:58:12 PM

Attachments (3)

Download all attachments as: .zip