HEPiX Fall 2005 at SLAC


Group Picture

Monday, 10 October 2005 (Collaboration Tools SIG)

Chuck Boeheim/Alan Silverman

Welcome to the Collaboration Tools SIG
Welcome, meeting logistics, and goals.
09:15-09:40 Christopher Huhn (GSI)
Wiki at GSI: Towards a collaborative application platform
As it is growing GSI's TWiki installation is evolving from a free-text documentation system to a rapid development toolbox for web-based groupware applications. I'd like to present some TWiki features and extensions used to enable this functionality and also give a report on the current installation status in general.
09:40-10:05 Achim Grindler (Karlsruhe)
WIKI based Change and Configuration Management Portal Solution
To operate a large IT configuration that should provide several high availability- and business critical services, it is necessary to pay attention to the IT Service Management processes Change- and Configuration management. To achieve a sound operation and high quality of service level for our customers, we had to change our classical technique driven behaviour to a more and more service- and customer oriented behaviour. I will present how the Department 'Services for Desktopsystems and Datanetworks' in Forschungszentrum Karlsruhe handles these IT processes. A locally developed WIKI based collaboration tool to support these tasks will also be shown.
10:00-10:30 Wolfgang Friebel (DESY)
FormFactory - Template based web forms
FormFactory is a work in progress to ease the building of web forms for non experts. It is based on state of the art perl modules. Main features are the separation of program flow (CGI::Application), layout (using CSS) and content (Templates). The form input can be validated and immediate feedback be given, the form output can be sent as mail, presented on screen, stored in a database (oracle, mysql, ...) etc. Simple applications can be created by just writing a config file while more advanced tasks require writing your own templates.
11:00-11:30 Thomas Baron (CERN)
Indico: An Event Management Software
The Integrated Digital Conference (Indico) software is a web-based application used for organising and scheduling events ranging from seminars to large conferences. The presentation will detail the history of the project, its main features, its possible extension as a live collaboration tool and its integration in the local conference room infrastructure.
11:30-12:00 Alf Wachsmann (SLAC)
Voice over IP with Skype
I will talk about the different features of Skype, their costs, and security problems.
12:00-12:30 John Gordon (CCLRC-RAL)
Access Grid - Support in the UK
Access Grid is a modern system of multipoint video conferencing designed mainly for dedicated conference rooms. Access Grid will be introduced and its use and support in the UK will be decribed.
13:30-14:00 Alexandre Lossent (CERN)
Collaborative tools in NICE
NICE is the Windows desktop environment at CERN. I will present some features of Microsoft products that help sharing information and documents.
14:00-14:30 Philipe Galvez (Caltech)
VRVS (Virtual Room Videoconferencing System) is a unique, globally scalable next-generation system for real-time collaboration by small workgroups, medium and large teams engaged in research, education and outreach. VRVS operates over an ensemble of national and international networks. Since it went into production service in early 1997, it has become a standard part of the toolset used daily by a large sector of HENP, and it is used increasingly by other DoE/NSF-supported programs. Today, the VRVS Web-based system is regularly accessed by more than 30,000 registered hosts running VRVS software in more than 120 countries. There are currently 83 VRVS "reflectors" that create the interconnections and manage the traffic flow, in the Americas, Europe and Asia. New reflectors installations have been performed in Serbia, Russia, Guatemala, etc. VRVS is global in scope: it covers the full range of existing and emerging protocols and the full range of client devices for collaboration, from mobile systems through desktops, to installations in large auditoria. A major architectural change is currently in development. The new VRVS generation is called EVO. EVO Beta Testing is expected to be started in the next coming weeks. It will be given a description of both the current operational state of VRVS service and also of the new architecture, including some of the new and advanced functionalities.
14:30-15:00 Steven M Goldfarb (Univ of Michegan/CERN)
The LCG RTAG on Collaborative Tools
Part 2
I report findings and recommendations from the LCG Project's Requirements and Technical Assessment Group (RTAG12) on Collaborative Tools for the LHC. A group comprising representatives of the LHC experiments, CERN IT and HR divisions (departments), and leading experts in the field of collaborative tools gathered information and analyzed the situation at CERN and member institutes of the LHC. Following a year of deliberation, they produced the report: CERN-LCG-PEB-2005-07, describing in detail the needs of the LHC, the current status, and making recommendations to bridge the gap between the two. The talk will summarize the findings and the recommendations, and pose some potential scenarii for implementation.
15:30-16:00 Brooks Collins (SLAC)
Remote Collaboration Working Group
16:00-16:30 Christian Helft (IN2P3)
Collaboration at IN2P3 and Future Deployments of Video Conferencing in HEP
The infrastructure deployed at IN2P3 to support videoconferencing will be described. A new phase including CERN participation has begun with the recent delivery of new equipment with promising new functionalities. Discussion on fitting this new equipment into other existing and future HEP wide collaboration infrastructure gives way to scenarios of further deployment of videoconferencing facilities for our community.
16:30-17:00 Alexandre Lossent (CERN)
Web hosting Services at CERN
CERN is providing a central web hosting infrastructure for more than 7000 web sites. I will review the status, recent achievements and expected evolution of this service.

Tuesday, 11 October 2005 (HEPiX)
09:00-09:25 Jonathan Dorfan (SLAC Director)
09:25-09:45 Multiple
Start of the HEPiX meeting. Introduction, status, goals...
09:45-10:00 Chuck Boeheim
SLAC Site Report
10:00-10:15 Martin Bly
RAL Site Report
10:15-10:30 Helge Meinhard
CERN Site Report
11:00-11:15 Alexander Withers
BNL/RHIC Site Report
11:15-11:30 Roberto Gomezel
INFN Site Report
11:30-11:45 Peter Gronbech
Oxford Site Report
11:45-12:00 Corrie Kost
TRIUMF Site Report
12:00-12:15 Jonathan Schaeffer
IN2P3 Site Report
12:15-12:30 Walter Schoen
GSI Site Report
tba Cary Whitney
NERSC Site Report
13:30-14:00 Robert Petkus (BNL)
OpenLDAP Configuration and Tuning in the Enterprise
OpenLDAP has been chosen as a directory service to replace NIS at the RHIC and USATLAS computing facilities. The advantages of OpenLDAP are discussed, as are various performance configurations and snags encountered along the way.
14:00-14:30 Hege Hansbakk (CERN)
Managing software licenses in a large organization
14:30-14:50 Ruben Gaspar Aparicio (CERN)
Progress report in CERN Certification authority deployment and Single Sign On with Certificates
Certificates can be used to authenticate on various applications, including websites, and can therefore provide some 'single sign on' mechanism, available on multiple platforms, in multiple browsers. Mapping existing Certificates (like Grid certificates) to CERN accounts as well as issuing new certificates to all CERN users is the goal of the CERN Certification authority project, as well as strengthening security by using Smartcards to store certificates.
14:50-15:00 John Gordon (RAL)
RAL Progress in Single Sign On
15:30-16:00 Richard Mount (SLAC)
Scientific Computing at SLAC
16:00-16:30 Bob Cowles (SLAC)
Security Update
Rainbow Crack Video
16:30-17:00 Marc Mengel (FNAL)
Fermilab Plone Update
How we've improved performance and services.
Conference Dinner at Computer History Museum - Starting at 7:00pm

Wednesday, 12 October 2005 (HEPiX)
09:00-09:15 Manfred Alef
GridKA Site Report
09:15-09:25 Pierrick Micout
DAPNIA Site Report
09:25-09:35 Michele Jouvin
LAL Site Report
09:35-09:45 Sabah Salih
Manchester Site Report
09:45-10:00 Kelvin Edwards
JLAB Site Report
10:00-10:15 Fraser R. Speirs
ScotGrid Site Report
10:15-10:30 Stephan Wiesand
DESY Site Report
11:00-11:30 Reinhard Baltrusch (DESY)
Experiences with WSUS/SUS as Patch Deployment Solution for Windows at DESY
Comparison of WSUS and SUS, advantages of WSUS for distributing patches for over 2500 Windows Systems.
11:30-12:00 Sean Roberts, Jean Pierre (SLAC)
Microsoft Clustering
12:00-12:30 Ruben Gaspar Aparicio (CERN)
Nice Admin towards running windows with no admin privileges
Nice Admin is an application that allows you to run certain processes as administrator using the same domain user account used to get a session running as least privilege. This application is part of the Non-Admin pilot project (http://winservices.web.cern.ch/WinServices/docs/NonAdmin/).
13:30-14:00 Tom Abel (SLAC/Kavli Institute for Particle Astrophysics)
Computational Cosmology
There is a now well calibrated standard model for the formation of structure in the universe. This offers a framework within one can conduct ab initio calculations of the formation of stars and galaxies, practically starting from the big bang. This talk will present results from such calculations, explain the relevant physics and illustrate the most challenging and interesting questions for the near future. The computational challenges will be highlighted in particular.
14:00-14:30 Alexander Withers (BNL)
Flocking at BNL: Implementing a General Use Queue
The RHIC/USATLAS Computing Facility has a large amount of computing resources at its disposal. Much of these resources are divided between the various experiments which has lead to inefficient use. With an ever growing dataset it has become critical to make efficient use of empty computing cycles. One of the goals in implementing Condor at the RACF is to provide for a general queue that can fill out the gaps in computing. This requires a technical solution in Condor that will allow for the utilization of these empty computing cycles while still respecting the boundaries between each of the experiments. This talk presents the possible solutions and issues for implementing a general queue that spans all available computing resources.
14:30-14:45 Tim Bell (CERN)
Activities following batch workshop in Spring HEPiX
During the HEPiX spring 2005, there was a workshop on batch systems and the grid. This talk will provide the status of some of the work in progress. Topics to be covered include community support for batch systems, use of the Glue schema, BLAHP and job progress monitoring.
14:45-15:00 Francesco P. Prelz
EGEE Batch System work
Approaches to batch system description, match-making and transfer of job control in the context of the EGEE project and the 'glite' software collection.
15:30-16:00 Troy Dawson (FNAL)
Update on Scientific Linux
16:00-16:30 Andras Horvath (CERN)
Linux status at CERN
The talk covers the migration status from CERN Linux 7 to Scientific Linux, current certification work and discusses the distribution of choice for the LHC startup in 2007.
16:30-17:00 Marc Mengel (FNAL)
Become Linux Distro Agnostic with Linux Standards Base
A possible solution for software distribution and GRID.
Scientific Linux BOF
A time for an interactive session so that the various labs can talk about how they can help, issues the S.L. developers have, and issues the labs have.

Thursday, 13 October 2005 (HEPiX)
9:00-9:15 Roger Jones (CERN)
Storage Task Force Report
The STF apologises for not being able to present a progress report at HEPiX today. Sudden changes in personal circumstances made this impossible. The discussions in the STF lead to many questions. Some of them we are still busy answering. Most notably hardware recommendations and predictions need to be fine tuned. Additionally our analysis made us address the developments of mass storage. We hope to able to circulate a draft of the final report by the end of the month to allow the HEPiX community to comment.
9:15-9:30 Gilbert Grosdidier (LAL-Orsay and CERN/LCG-GD)
The Light-Weight Disk Pool Manager
This tool is being developed within the LCG and EGEE framework, to replace the Classic SE.
09:30-10:00 Robert Petkus (BNL)
Panasas at the RCF
Panasas, an object oriented, distributed filesystem and hardware/software storage solution has been in production at the RCF for the past year. Basic architectural design, performance, experiences, and alternate solutions are discussed.
10:00-10:30 Andrew Hanushevsky (SLAC)/Jean-Yves Nief (IN2P3)
Xrootd - Present and Future
Xrootd at IN2P3
Xrootd provides scalable high-speed fault-tolerant access to Babar data. This talk will show what these terms really mean and where xrootd is going in the HEP community.
11:00-11:30 Jean-Yves Nief (IN2P3)
BaBar data distribution using the Storage Resource Broker
The Storage Resource Broker is middleware providing an uniform interface to heterogeneous storage systems. In this talk, we will present how it is being used within BaBar to make data distribution between SLAC and CC-IN2P3. Prospects to include RAL to the system will be also shown.
This is a joint effort of Wilko Kroeger (SLAC), Adil Hasan (CCLRC-RAL) and myself.
Additionally, the usage of the Storage Resource Broker by various experiments at CC-IN2P3 will be presented.
11:30-12:00 Radovan Chytracek (CERN)
Databases for physics, their deployment and monitoring
12:00-12:30 Zhenping Liu (BNL)
USATLAS dCache system at BNL
A large distributed grid-enabled storage system, dCache, has been running in production at BNL for about one year. We will talk about status, experiences, issues and plans of the system. Our experiences about service challenge activities at BNL will also be presented.
13:30-14:00 Matteo Melani (SLAC)
Yet another Grid Project: the Open Science Grid (OSG) at SLAC
Since July 22nd the OSG Grid has been open for business. Thousands of CPUs and Terabytes of disk space are now available to the computational intensive sciences and in particular to the HEP community. But what is the Open Science Grid really? What services does it offer? How does its software stack look like? SLAC is contributing to the OSG effort by providing computing resources, designing new services and defining security policies. In this talk we will present the OSG in a nutshell and talk about the SLAC experience in running the OSG software stack.
14:00-14:30 Miroslav Siket(CERN)
Update on Lemon
An update on the current development of Lemon will be given along with the feedback received on usage and installation from other institutes.
14:30-15:00 David Kelsey (RAL)
Planning for LCG emergencies
We (JSPG) have recently started work on plans to help LCG run during or after emergencies (security related or other problems). We welcome feedback from HEPiX.
15:30-16:00 Keith Chadwick (FNAL)
FermiGrid - Status and Plans
FermiGrid is a cooperative project across the Fermilab Computing Division and its stakeholders which includes the following 4 key components: Centrally Managed & Supported Common Grid Services, Stakeholder Bilateral Interoperability, Development of OSG Interfaces for Fermilab and Exposure of the Permanent Storage System. The initial goals, current status and future plans for FermiGrid will be presented.
16:00-16:30 Fraser R. Speirs (ScotGrid/University of Glasgow)
Evolving Local Resources Towards the Grid
The next challenge for grid deployers is to grid-enable local resources and evangelise local users to adopt grid. We present the past and future of ScotGrid.
16:30-16:45 Michel Jouvin (IN2P3/LAL)
GRIF Project

Friday, 14 October 2005 (Infrastructure/Machine Room Issues SIG)
09:00-9:30 Hugo Cacote (CERN)
Testing High Performance Tape Drives at CERN
Current and future high-performance tape drives available on the market will support data transfers at rates exceeding 1 Gbit/s. The Hierarchical Storage Management System used at CERN is based on an infrastructure composed of Linux file servers that transfer data over a Gigabit Ethernet backbone to Linux tape servers. Today each of these is connected to a single tape drive via fibre channel, using a point-to-point topology. Unfortunately, this architecture won't be able to work efficiently above the Gbit/s device rate. In this presentation we show several possible enhancements to the current architecture aimed at easing the efficient integration of high-performance tape devices. The impact on performance of using high memory capacity IA64 tape servers and the possibility to run such tape drive hosts with local disks buffering incoming data streams are some of the issues addressed. The preliminary results of setting up a server dedicated to media migration, running all the HSM software components and attached to high-speed drives such as LTO-3, are also presented. Finally we also evaluate the effect on performance of the different components of tape labels and assess how to efficiently process them or perhaps even omit them in the future.
09:30-10:00 Helge Meinhard (CERN)
Computer Hardware and Procurement at CERN
CERN as an international organisation needs to follow strict rules for all procurement procedures. These rules are not always well adapted to procurements of computing equipment, the market of which undergoes rapid technical and commercial evolution. I'll describe the constraints, how we react to them, will give an indication of the timescales involved, and then move on to describing a few of our recent or ongoing procurements in detail. After showing briefly a few measurements of power consumption, I'll conclude with some observations and issues.
10:00-10:30 Manfred Alef (Karlsruhe)
CPU Benchmarking at GridKa
The price of power and cooling is an important aspect when building large PC clusters. Dual-Core CPUs as well as mobile CPUs may offer an improved computing power per watt. A comparison between several CPUs with respect to computing power vs. electric power consumption vs. heat dissipation will be presented.
11:00-11:30 John Gordon (RAL)
Accounting at RAL
11:30-12:00 Chuck Boeheim (SLAC)
An Incident with a 200ft Tree
Last May, a large tree fell across the power lines into SLAC, providing an unanticipated failure failure mode for all systems. The response provided a number of communication challeges, and a number of lessons that may (or may not) help the next unanticipated challenge.
12:00-12:30 Tony Cass (CERN)
The CERN Computer Centres
An overview of the readiness of the Computer Centre for LHC, including an overall vision for Quator, Lemon, and Leaf.
12:30 Adjourn