Docstoc

cern-intas-03-52-4297-finalreport

Document Sample
cern-intas-03-52-4297-finalreport Powered By Docstoc
					                                                                                                Ref. No 03-52-4297




                             Final Report on
               RESEARCH PROJECT AND NETWORKING GRANTS
TECHNICAL INFORMATION
Call       CERN-INTAS Call 2003
specification:
Title:              Advanced algorithms and tools for LHC Computing GRID
Reference no:       CERN-INTAS 03-52-4297
Total grant:        120000 Euro
Start date:         01.07.2004
Duration:           24 months
Co-ordinator:       Jürgen Knobloch, CERN-IT, CH-1211, Geneve 23, Switzerland,
                    juergen.knobloch@cern.ch
Team                Jürgen Knobloch, CERN-IT, CH-1211, Geneve 23, Switzerland,
information:
                    juergen.knobloch@cern.ch
                    Mirco Mazzucato, INFN, Padova, Italy, mirco.mazzucato@pd.infn.it
                    Marcel Kunze, FZK, Karlsruhe, Germany, Marcel.Kunze@iwr.fzk.de
                    Alexander Kryukov, SINP MSU, Moscow, Russia, kryukov@theory.sinp.msu.ru
                    Yuri Tikhonov, BINP, Novosibirsk, Russia, tikhonov@inp.nsk.su
                    Andrey Minaenko, IHEP, Protvino, Russia, minaenko@ihep.ru
                    Vladimir Gavrilov, ITEP, Moscow, Russia, gavrilov@mail.cern.ch
                    Vladimir Korenkov, JINR, Dubna, Russia, korenkov@cv.jinr.ru
                    Yuri Ryabov, PNPI, Gatchina, Russia, ryabov@pnpi.nw.ru
Final report due:   End of the second year of the project
Submission date     18.01.2007
of this report:


1       RESEARCH
1.1. Overview of Research Activities / Conformance with the Work Programme (1/2-1 page)
     The work was planned in the flat time scheduling over all tasks. The goals of the project have been achieved,
basically in accordance with the Work Programme.
    In July 2005 the presentation has been given by A. Kryukov, leader of the SINP MSU team, on the CERN-INTAS
workshop (11-12 July 2005, CERN), where he summarized the results obtained by the consortium during the first year
of this project (http://theory/dokuwiki/intas/cern-intas03/papers).
     The final meeting of the project was held on 25 June 2006 in Dubna, where J. Knobloch, the project Coordinator,
Les Robertson (both from CERN IT team), all Russia team leaders and key participants from Russia teams
participated. There were presented reports by the task leaders and key participants (see web-page of this meeting
http://lcg.jinr.ru/section.asp?sd_id=84).
     There were no serious deviations form the Work Programme. For sure, the intense development in this field, grid
technologies for science and especially for High Energy Physics, has that natural subsequence as need of corrections
concerning the methods and tools we have planned to use in our work. Then, our project was related closely to the
LHC Computing Grid project (http://lcg.web.cern.ch/LCG/), as formulated in the Work Programme, and we have
had to take into account new solutions taken in this ―mother‖ project.

                                        CERN-INTAS 03-52-4297 Final Report

                                                     Page 110
                                                                                                      Ref. No 03-52-4297



     In the report we emphasize on main results obtained in the tasks of the Work Programme, where different teams
participated:
  T1. Advanced mathematical techniques for LCG (task leader A. Kryukov, collaborating teams SINP-MSU, ITEP and
INFN-Padova);
  T2. GRID monitoring tools (task leader Yu. Ryabov, collaborating teams PNPI, SINP, BINP, IHEP, ITEP, JINR);
  T3. Optimization of data access and transfer in the LCG (task leader V. Korenkov, collaborating teams JINR, BINP,
IHEP, CERN-IT, FZK);
 T4. The development of test suite for new LCG middleware (task leader I. Bird, deputy task leader E. Slabospitskaya,
collaborating teams IHEP, CERN-IT, ITEP, JINR, PNPI);
  T5. The use of the Windows platform for LCG tasks (task leader M. Kunze, deputy V. Korenkov, collaborating teams
FZK and JINR);
 T6. Deployment of LCG infrastructure for creating Russian Grid segment (task leader V. Ilyin, collaborating teams
CERN-IT, SINP-MSU, BINP, IHEP, ITEP, JINR, PNPI).
One should note, that in the subtask T6.1
     T6.1. Installing and testing components of the LCG infrastructure to achieve full-scale functionality
all Russian teams participated in collaboration with CERN-IT - as it stated in the Work Programme, this subtask has
infrastructure character for the project in whole, because other tasks were planned to work out by use the experience of
Russian teams in the LCG deployment. Here we have achieved serious development – Russia LCG sites (most of them
are partners of this INTAS project) are creating the Tier2 cluster with basic functionality at the Tier2 level with
fraction of the Tier1 functionality (e.g., transferring and storage part of ESD an RAW data from Tier1s to Russia,
storage of MC data simulated in Russia and providing access of WLCG users to these data). This activity has got
systematic development in Russia according to the LCG plans on creating the global grid infrastructure of regional
centers for LHC data elaborating, storage and for distributed analysis. The corresponding reports were done by V. Ilyin
(SINP-MSU team) on regular official CERN-Russia meetings twice per year in 2004-2006, in spring and autumn (see
presentation by V. Ilyin on the final meeting for this CERN-INTAS project).

1.2. Scientific Results (max. 2 pages, + list of references)
Below short formulation of the results obtained in the 1 st year are given, and desciptions of the results obtained in 2 nd
year, task by task. The results of 2nd year have been reported by site representatives at the final project meeting held in
JINR (Dubna, Russia) on 25 June 2006 (http://lcg.jinr.ru/section.asp?sd_id=84).
T1. Advanced mathematical techniques for LCG (leader A. Kryukov, teams SINP-MSU, ITEP and INFN-Padova)
    1st year. It was created a set of software scripts for transferring InputSandbox to WN (Working Nodes) directly
(bypassing RB – Resource Broker). The study has been done to find ways to improve the RB performance and
reliability of job submission and plan for further research have been elaborated.
   Massive MC simulation tasks asked intensive data exchange between WNs and SE for ALICE, CMS and LHCb
were studied at LAN environment based on OpenPBS batch system - dataflow bottleneck was located and application
SW was adjusted. In order to increase efficiency of job execution by means of killing of looping jobs special
monitoring scripts were tested output files on WN local disks to detect and kill stalled jobs.
   2nd year. For improving the job submission in LCG there was elaborated the improved LCG job flow architecture
and API for set of LCG commands: cg-job-submit, local-job-wrapper, lcg-se-clear, lcg-se-test, TRANSMIT_SE.
Developed approach allows upload and improve robustness and productivity of LCG RB without modification in
present LCG components, by use the standard grid protocols only and by use the standard grid security infrastructure.
It opens ways to realize job buffering to reduce wall-time of job submission and use single copy of executables for
multiple submission, http://theory/dokuwiki/intas/cern-intas03/papers.
   There was developed data flow optimization in realistic DC environment ALICE and LHCb MC production. Within
the massive LHCb DC there were obtained significant increase (by 30-40%) of the LHCb job submission rate for CPU
with hyperthreading. During the massive ALICE DC sessions were located the interference with LHCb jobs resulted
on often crashes of NFS. It was proposed the optimization of job submission interference between two (and more)
experiments by restriction of the experiment queues on local resources to 10 simultaneous jobs and by optimizing the
communication with disk server. The simulation of intensive data flow during data analysis was performed by use of
CMS-like job analyses of 1000 events in 100 seconds.
T2. GRID monitoring tools (leader - Yu. Ryabov, teams PNPI, SINP, BINP, IHEP, ITEP, JINR).


                                          CERN-INTAS 03-52-4297 Final Report

                                                        Page 210
                                                                                                     Ref. No 03-52-4297



   1st year. An application monitoring system 'apm' was created allowing grid user to track the execution process of his
job via a web-interface. The monitoring system was deployed at SINP MSU and tested by IHEP and BINP teams. The
application package CompHEP adapted for use through created grid infrastructure was used for these purposes.
    Database was filled with real data from LCG testbed monitoring. Two database backends (MySQL and Orcale)
were tested. The corresponding statistical algorithms were developed. Working prototype of the monitoring system
was installed at PNPI and performs monitoring of LCG2 testbed with 15 min polling interval. Services SW for grid
monitoring and accounting tools were installed and tested at ITEP LCG segment.
   2nd year. For development of monitoring tools that produce integrated characteristics by use of statistical information
gathered from various Grid elements the complete software suite for statistical monitoring which will include the
monitoring agents, database, statistical analysis program and several front-ends (web interface, APIs) were created and
tested. This statistical information could be used by Grid end-users and resource brokers to qualify the elements by
durability and accessibility for Grid applications as well as for a long-term estimation of the resource load. During the
development period a substantial changes in the Grid middleware had happen. This forced to rewriting several
fragments of code and increased debugging time. Final system consists of six basic components: 1) Database (DB),
holds all information related to monitoring and resource classification, MySQL is used as a DBMS but it may be
replaced by any relational DBMS like Postgres or Oracle; 2) MDS Data Collector (MDC), this tool takes resource state
information from Grid information system (MDS) and stores it in the database; 3) RGMA Data Collector (RDC), this
tool takes resource testing information from RGMA and stores it in the database; 4) Data Processor (DP), this tool
takes raw information from the database, processes it and stores the results back; 5) Service Web Interface (SI), this
component allows an administrator to examine monitoring information stored in the database in a visual form; 6) User
Web Interface (UI), this component in intended to be installed on a Grid UI node and allows user to work with the
system from a Web browser. It shows classified resources as well as their parameters and makes it possible to submit
jobs to selected resources. Other usual tasks like proxy-certificate management, job status monitoring, job output
retrieval and so on are supported as well.
   For monitoring of application jobs it was found that existing LCG job-monitoring system offers information only
about a few discrete states of a job in grid (e.g., submitted, active, done) and output and error messages can be obtained
only after completion of job execution. As a result of the research performed an extension of existing LCG job-
monitoring system was suggested to allow users to have intermediate job output (through time intervals specified by
users ) for detailed job execution control. Basic features developed tools are: no LCG MW modification required
(wrappers + additional server), access to the intermediate job output via Web-interface, authorization is based on the
standard GSI certificates and proxy certificates. Starting Web-page for interested users (with instructions) is
http://grid.sinp.msu.ru/acgi-bin/welcome.cgi. In addition user has a possibility to add castom filter which
permit to select from full stdout/stderr flow the interesting information - the user just provides a script/binary in the
format: stdout-filer-stdout.

T3. Optimization of data access and transfer in the LCG (leader V. Korenkov, teams J INR, BINP, IHEP, CERN-IT, FZK).
  1st year. CASTOR and dCache systems have been installed for supporting the data storage in the LCG environment;
DPM system installation is underway (JINR, IHEP) and tested with participation of BINP. The JINR and IHEP
participants took part in the development and debugging of the new CASTOR2 version at CERN, in particular, new
Garbage Collector was created, the Distributed Logging Facility (DLF) has been improved and new DLF server is
installed for testing, tools for monitoring of free and occupied space of data storage have been developed.
  2nd year. LCG-2 and gLite middleware were successfully deployed in all Russian sites participated in this project.
Communication links in Russia and to GEANT has been greatly improved. Monitoring services and tools for collecting
statistics off internal resources in Russia have been installed. The storage software components (solutions) have been
studied - CASTOR, DPM and dCache SE types, SRM protocol. It was investigated efficiency of RLS and LFC for data
location API and tools in RDIG environment. The VOs schemes for LCG Service Challenges (SC3 and SC4), Data
Chellenges and analyze data management framework were tested. In 2006 the total storage space in RDIG sites
reached several tens TB and monitoring tools for more precision investigation of these storage resources and links
quality has been developed for aggregation of RDIG data storage resources in terms of Russian distributed Tier2
Cluster model. The dCache software was choose as base for aggregation of the data storage resources for Russian
distributed Tier2 Cluster model.
   There was investigated the job scheduling problem in CASTOR2 system. The straightforward way is to use the LSF
(Load Sharing Facility), which has well known problems, like scheduling is complicated, scheduling takes a time, -
each running job involves 5 processes etc. In 2003 an alternative way had been proposed to schedule jobs in CASTOR
by MAUI, an open source package. MAUI is just a job scheduler (not a Job Management System) with WiKi (plain

                                         CERN-INTAS 03-52-4297 Final Report

                                                       Page 310
                                                                                                    Ref. No 03-52-4297



text) interface, which allows easy integration in CASTOR (via rmmaster). MAUI has been developed further, and
Cluster resourses now offers a "full fledged" Job Management System, called MOAB, with advanced job scheduling
based on MAUI, which scales up to larger number of jobes and nodes.To fully exploit the advanced features of MAUI
(or MOAB) corresponding changes to rmmaster have been applied. As a result there was investigated the possibility to
use the open source ORACLE XE software for data management in CASTOR2. Based on the code of DPM (LCG Disk
Pool Manager) there was created module CASTOR DSI (Data Storage Interface) for gridFTP v2 (Globus Tooklit 4.x).
There has been started the work on implementation of gsiftp (Grid Security Infrastructure File Transfer Protocol) in the
СASTOR2 architecture.

T4. The development of test suite for new LCG middleware (leader I. Bird, deputy E. Slabospitskaya, teams IHEP,
CERN-IT, ITEP, JINR, PNPI)
  1st year. There were created test suites for Storage Resource Manager (SRM) v1.1 and Grid File Access Library
(GFAL), fully integrated into the TSTG testing framework. The Site Functional Test (SFT) was developed and ported
for the gLite environment. The I/O storm test suite has been developed for gLite middleware to test the gLite I/O
server. The test suite was integrated into the TSTG testing framework. The LCG-2 test suit was assembled in the ITEP
in the following configuration: LCFG, CE, SE (2 TB), UI, WN (32 nodes), Monitoring server, User support server.
   2nd year. Within the common problem of porting the EDG/TSTG framework to the EGEE middleware components
(important for testing of production grid sites) and to provide daily testing the grid specific software right after the
build, the NMI system has been ported as first step. One of the goals was to submit and execute tests from edg-tests
framework, used to test grid middleware in CERN IT\GD group. This method was implemented but was found as not
effective. Then, new scheme was probed for this purposes - SAM (Service Availability Monitoring) framework, and
the integration of TSTG framework to the SAM has been formulated for further realization. The development of
SRM/GFAL test suites has been created for new versions of these services. The specialized testbed for developing test
suites was created in IHEP and used for developing new tools for test launching using HP SmartFrog Configuration
Framework tool.

T5. The use of the Windows platform for LCG tasks (leader M. Kunze, deputy V. Korenkov, teams FZK and JINR)
   1st year. The VMware software has been studied as a tool for creation of clusters for distributed computations and
batch farms for deployment in the LCG Tier2 level. The computing farm of 50 computers with Windows2000 were
mobilized in the Dubna-Grid system as a basic system has been used for loading the virtual (guest) computers with the
Scientific Linux operating system. The first results shown reasonably high effectiveness of this solution. The complex
facilities have been used for testing run of simulation jobs for the CMS experiment.
   The possibilities for improvement of LCG portal functionalities based on the web-services in the framework of
.NET platform have been studied, in particular for monitoring the disk and processor resources usage. The software has
been deployed at the Russian LCG portal for providing services as alfa-version. The first tests and results have shown
good opportunity to construct a system in the heterogeneous environment with data exchange by means of web
services technology.
   2nd year. The VMware testbed in the Cluster Dubna-Grid was developed to about 200 WS w/ Windows + 3 servers
under the SLC306 as the guest OS under VMware WS and Vmplayer. There are two clusters now: production and
development. Guest OS (SLC306) was installed by use ―Warewulf Cluster Project‖: minimal part of the OS has to
bootstrap from the server via PXE. The rest of the OS and user utilities exported from AFS to the usual place via
mount --bind. For getting persistence storage area real disk partitions were exported from VMware to the guest OS.
Maui and Torque installed for batch job processing. Batch system configured for accepting all jobs with corresponding
parameters. AFS and dCache tuned to work as data and software storage. Resulting cluster was used for real work as
provision resources for LCG at JINR. For further development - shorten bootstrap and switch to root FS on AFS,
deployment of gLite WN in AFS with on-the-fly configuration at boot time, development of VLAN based scheme for
grow cluster from different locations, accommodation of more resources in the cluster
   The prototype of LCG portal based on the web-services in the framework of .NET platform was realized with
Monalisa server (Linux based platform), client based on Windows.NET platform, SQL server and user's Web interface.
The prototype was tested in work with RDIG sites through the Web interface created.

T6. Deployment of LCG infrastructure for creating Russian Grid segment (leader V. Ilyin, teams CERN-IT, SINP-
MSU, BINP, IHEP, ITEP, JINR, PNPI).




                                         CERN-INTAS 03-52-4297 Final Report

                                                      Page 410
                                                                                                        Ref. No 03-52-4297



    The subtask ―T6.1. Installing and testing components of the LCG infrastructure to achieve full-scale functionality”
has infrastructure status in the project and has value for all other tasks, as explained in the section 1.1 of this Report,
were main results are formulated. Some specific results in this stream are formulated below.
     The Core Infrastructure Center (LCG/EGEE CIC) has been created in SINP-MSU with some functions distributed
at JINR and ITEP – thus distributed model for CIC has been elaborated with full set of basic grid services
implemented. From May 2005 the Russian CIC has started the weekly on-duty operation shifts, rotating with four other
European CICs – CERN, in UK (RAL), France (CC-IN2P3) and Italy (CNAF) http://egee.sinp.msu.ru.
  The Regional Operational Center (LCG/EGEE ROC) has been created in IHEP (http://www.ihep.su/egee),
with distribution of some functions between JINR (http://rocmon.jinr.ru:8080/), ITEP (http://egee.itep.ru),
PNPI (http://egee.pnpi.nw.ru) and SINP-MSU http://egee.sinp.msu.ru. The Russian ROC successfully
started the operation of the RDIG sites as a segment in the LCG/EGEE infrastructure, providing several Data and
Service Challenges of the LHC Experiments (ATLAS, ALICE, CMS and LHCb). The User Call Center has been
created by ITEP team in collaboration with FZK ( http://ussup.itep.ru).
  In testing of the distributed model of the CIC and ROC all Russian teams participated, including efforts of
participants from BINP, in particular, taking part in the upgrade of the CERN grid Certification Authority
infrastructure.
   This infrastructure work was done basically by financial support of EU EGEE project (except of BINP team) and
from different Russian sources. However, the corresponding projects are of pure infrastructure character without
research component. As the CIC and ROC are created in Russia in distributed model, seriously differed from CICs and
ROCs in other European centers of the LCG/EGEE infrastructure, the corresponding research had to be made. So, the
main efforts supported by this CERN-INTAS grant were applied to study of possible distributed models, in particular
to minimize risks of the operating this new type infrastructure in local conditions. This study is planned to continue in
further years toward the start of LHC operation in 2008, in particular, for creating in Russia the next generation of joint
CIC and ROC (ROC in terminology of EGEE-II project – http://grid.sinp.msu.ru/grid/roc/main (in Russian)),
again by use of the distributed model, and taking in account specifics of the deployment of new MW (gLite
components).
    As to subtasks T6.2 and T6.3 the following results were obtained.
  1st year. The main aim of the subtask T6.2 was studying, installation, testing and evaluation of the middleware on the
basis of web-services for creation of grid-systems with the Open Grid Services Architecture (OGSA) for distributed
calculations and data processing. The work was carried out along three directions: (1) obtaining an operational
experience with the middleware toolkits and creation on their basis new or modified grid-services; (2) installation of
testbeds for quantitative studying middleware components; (3) measurements of performance characteristics of high
level services (in particular, workload management system, services of file transfers, information service and others).
The purpose of testing was the evaluation of functional abilities, productivity, scalability, reliability, compatibility with
other software, difficulties at installation process, configuration and administration, convenience and simplicity of use
of high level services. For these aims, testbeds in SINP MSU, JINR and CERN have been created.
    In addition to the comprehensive investigation of the Globus Toolkit 3 executed at the first stage (2004), base
components of following open source toolkits and middleware packages have been studied, installed and tested:
Globus Toolkit 4 (GT4), OMII, and selected gLite components. The work was carried out in close collaboration with
the LCG/GTA group at CERN.
    The results obtaibed in siubtask T6.3 the two versions of Russian LCG portal have been developed
(http://www.egee-rdig.ru). The first version aims to collect the information about computing resources of the
RDIG sites and to provide the collaborating sites with a possibility of computing resources planning via the web-
interface. The second version of the RDIG/LCG portal have been designed with improvements by including the
additional services. These services are based on special information system for monitoring of Russian LCG sites
    2nd year, subtask T6.2. Testing/evaluation of OMII MW was performed with goal to view potential use of some its
components for EGEE/LCG MW. The following functionality provided by the OMII MW was viewed: installation and
configuration of the MW on different platforms, testing of performance, concurrency, reliability, usability and
management of the services, Job Services, Data Services, security, adding an application, adding a new service,
Performance and concurrency of dummy services, interoperability with gLite Workload Management System (WMS).‫‏‬
    OMII evaluation results: OMII is oriented towards web services rather than Grid architecture; there are no
Workload Management System & Resource Broker, Information System, Logging and Bookkeeping, Replica Catalogs
and Replica Management System. OMII does not support GSI and proxy credentials and Virtual Organizations

                                          CERN-INTAS 03-52-4297 Final Report

                                                         Page 510
                                                                                                 Ref. No 03-52-4297



management. There is no interoperability with Globus Toolkit and the gLite WMS. More details on OMII evaluation:

http://lit.jinr.ru/Reports/annual-
report05/omni-55.pdf.
    There was performed the Globus Toolkit 4 (GT4) testing for which the GT4.0.0 version was installed and properly
configured on a number of platforms (in total on 7 machines): Fedora Core 3, Redhat 9, SLC3 (Scientific Linux Cern
Release 3.0.4), SL4 (Scientific Linux Release 4.0).
    In this testing the WSRF implementations were compared: GT4-Java (Java WS Core of GT4); GT4-C (C WS Core
of GT4); pyGridWare (Python WS Core); the Perl-based WSRF::Lite. Also there was done functional comparison:
significant commonalities are seen with regard to dispatching and SOAP processing techniques. There was estimated
the overhead expenses of GT4 Java container: overhead expenses of GT4 Java container (i.e. first part which is
common for invocation of all services) and client application, estimation of average times for common WSRF
operations, average times for common WSRF operations (in GT4 Java container) on server side and client side and
investigation the influence of the number of created resources on the request time, estimation of GT4 Java container
scalability and robustness, GT4 Java container scalability = maximum possible number of parallel invocations before
server failure). There was done GridFTP and RFT testing.
    The gLite – GT4 WorkSpace Service (WSS) compatibility was studied. The goal was to verify compatibility of
WSS with gLite and evaluate the service. Results: succeeded in combining gLite+WSS, at the time of the study (Fall
of 2005) WSS proved to be too immature and unstable to be considered as real candidate for gLite component.
     There was done comparative studying: GT4 – gLite. In each main area gLite has services, which are absent in
GT4, but are substantial: security; virtual organization support (VOMS gLite); Execution Management - Workload
Management System; Data Management - support of storage resource internal organization: SE, SRM, file catalogues.
R-GMA (gLite) vs. MDS4: essentially different approaches.
   The gLite component testing was done: Fireman (file + replica) and Metadata catalogs

http://lcg.web.cern.ch/lcg/PEB/arda/public
_docs/CHEP04/CaseStudies/Fireman1.do
c  ; IO-server


http://lcg.web.cern.ch/lcg/PEB/arda/public
_docs/CHEP04/CaseStudies/gLiteIO%20-
%20testing-v2.pdf;                                  Workload Management System


http://lcg.web.cern.ch/lcg/activities/arda/p
ublic_docs/2005/Q1/ARDAatTaiwan-
apr05.ppt                 ; AMGA Metadata Service‫‏‬
  This work (OMII and GT4e testing/evaluation) was carried out in ARDA framework (partially supported by this
INTAS-CERN project) and in close collaboration with EGEE JRA1 team.

                                       CERN-INTAS 03-52-4297 Final Report

                                                     Page 610
                                                                                                   Ref. No 03-52-4297



   2nd year, subtask T6.3. The Russia LCG Web site version 2 was completed, tested and started the daily providing
the LCG users. New features: organize web based user support, provide more detailed monitoring and statistic
                                           ,
information (based on MonALISA system.)‫ ‏‬new graphical design and new information concept, integration with
monitoring system MonALISA throw web services .

List of references on publications and presentations related to the project.
Baud J-Ph. , Sekera Z., Lazin Yu., Motyakov V., Petoukhov V., Slabospitskaya E. ―SRM Testing for LCG-2‖. Talk on
International Conference ―Distributed Computing and Grid Technologies in science and Education‖, Dubna, 2004, 29
June-2 July. http://lit.jinr.ru/grid2004/program.htm#A0
V.A. Ilyin, A.P. Kryukov, A.P. Demichev and L.V. Shamardin, «Grid infrastructure RDIG — Center of basic grid
service in SINP MSU», Proceedings of the Conference «Science service in Internet» (Novorossiisk, Russia, September
2005), pp. 27-28, MSU, 2006, in Russian.
A. Kryukov and L. Shamardin, ―Implementation of vertex form factors in grid-oriented version of CompHEP‖, Nucl.
Instrum. Methods , vol.559, Issue 1, 2006, pp. 253-256.
V. Pose and B. Koblitz, "Testing of gLite-Amga Metadata Service", Proc. of 2nd International Conference «Distributed
computations and grid-technologies in science and education» (Grid2006, June 26-30, 2006, Dubna, Russia), Dubna,
June, 2006, ISBN 5-9530-0138-X, pp. 128-133.
A. Kiryanov. ―High availability for core LCG services‖, Book of Abstracts of the 2nd International Conference
―Distributed computations and grid-technologies in scence and education» (Grid2006, June 26-30, 2006, Dubna,
Russia), Dubna, 2006, p.74.
Petoukhov V., Sapunov M., Slabospitskaya E. ―SRM and GFAL Testing for LCG-2‖. Poster presentation on
International Conference CHEP04 (Interlaken, Switzerland), 2004, 27 Sept.-1 Oct.2004
http://indico.cern.ch/contributionDisplay.py?contribId=358&sessionId=23&confId=0
V. Korenkov, ‖JINR Participation in the LCG and EGEE Projects‖, Proc. Int. Symposium NEC'2005 (12-18
September, 2005, Varna, Bulgaria), pp. 170-173, Varna, 2006.
Y. Lyublev, A. Selivanov, M. Sokolov. ‖Analysis System of ITEP Network Links‖. Proc. Int. Symposium NEC'2005
(12-18 September, 2005, Varna, Bulgaria), pp.195-197, Varna, 2006.
I. Korolko. ―DIRAC - LHCb System for Distributed MC Production and Data Analysis‖, Proc. Int. Symposium
NEC'2005 (12-18 September, 2005, Varna, Bulgaria), pp.174-179, Varna, 2006.
Victor Pose et al. ‖OMII distribution evaluation activity‖. Presentation on training session prior the 3rd EGEE
Conference (Athens, 18.04-22.04.2005).
E. Slabospitskaya, G. Grosdidier and V. Petoukhov. ―gLite I/O storm testing in EDG-LCG Framework‖. Presentation
on Int. Symposium NEC'2005 (12-18 September, 2005, Varna, Bulgaria),
http://www.jinr.ru/unixinfo/temp_doc/NEC2005/slabospit_gliteIO.ppt
V. Jiltsov and V. Korenkov, "JINR and Russia in LCG Service Challenge Activity". Presentation on Int. Symposium
NEC'2005 (12-18 September, 2005, Varna, Bulgaria),
http://www.jinr.ru/unixinfo/temp_doc/NEC2005/Jiltsov-SC3-NEC-Slides.ppt
A.Demichev, V.Ilyin, A.Kryukov, L.Shamardin, «Federation of grid segments with different infrastructures».
Proceeings of 6-th Conference "Electronic libraries: future methods and technologies, electronic collections"
(RCDL’2004, Pustchino, Russia, October 2004), pp.33-37. in Russian.
http://www.impb.ru/~rcdl2004/cgi/get_paper_pdf.cgi?pid=22
Vl.V.Voevodin, A.P.Demichev, S.A.Zhumatii, V.A.Ilyin, A.P.Kryukov, «Use of technologies for distributed
operations with data in scientific research and education at Moscow university », Proceedings of the Conference
"Scientific service in Internet " (Novorossiisk, Russia, September 2004), pp. 258-259, MSU, 2004, in Russian.
A.P. Demichev and A.P. Kryukov, "Creation and functioning of virtual organizations in the RDIG infrastructure ".
Proceedings of the Conference "Scientific service in Internet " (Novorossiisk, Russia, 19-24 September 2005), pp. 8-9,
MSU, 2005, in Russian.
V.A. Ilyin, A.P. Kryukov, A.P. Demichev and L.V. Shamardin, «Experience with the EGEE Center of basic grid
services in SINP MSU», Proc. of 2nd International conference «Distributed computations and grid-technologies in
science and education» (Grid2006, June 26-30, 2006, Dubna, Russia), pp. 57–61, JINR, 2006, ISBN 5-9530-0138-X.

                                        CERN-INTAS 03-52-4297 Final Report

                                                      Page 710
                                                                                                     Ref. No 03-52-4297



A.P. Kryukov, A.P. Demichev, v.N. Kovalenko and V.V. Mitsyn, «Transferring of RDIG infrastructure on gLite
middleware», Proceedings of 2nd International conference «Distributed computations and grid-technologies in science
and education» (Grid2006, June 26-30, 2006, Dubna, Russia), pp. 72–76, JINR, 2006, , ISBN 5-9530-0138-X.
V. Kotlyar, Yu. Lazin, V. Motyakov, E. Slabospitskaya, V. Urazmetov. ―The specifity of using afs, quattor,castor when
working nodes are placed into virtual subnetwork‖, Proc. 2nd International Conference ―Distributed computations and
grid-technologies in scence and education» (Grid2006, June 26-30, 2006, Dubna, Russia), Dubna, 2006, ISBN 5-9530-
0138-X.
M. Sapunov. ―Using the EDG-tests framework via NMI‖. Presentation on 2nd International Conference ―Distributed
computations and grid-technologies in scence and education» (Grid2006, June 26-30, 2006, Dubna, Russia).
Yu. Lazin. ―Integration and adaptation of various hardware platforms to grid-environment based on LCG* and SGE**
software‖. Book of Abstracts of the 2nd International Conference ―Distributed computations and grid-technologies in
scence and education» (Grid2006, June 26-30, 2006, Dubna, Russia), p.99.
    Which scientific papers, presentations & patents have resulted directly from this project?
Summarize the scientific output (number of papers, etc.) in the table below:
                                          ALL PUBLICATIONS                               ONLY: Jointly by INTAS
                                                                                         and NIS Project teams
Scientific Output                         Published     In press /accepted   submitted
Paper in an International Journal         1
Paper in a National Journal (Russian)
Proceedings (conferences, workshops)      13                                             1
Book, Monograph
Internal Report                           6                                              2
Thesis (MSc, PhD, etc.)
Patent


1.3. Impact and Applications
 The work of Russia teams was performed in close contact with CERN IT, INFN-Padova and FZK teams within the
LCG project. Thus the results of this work are already in use by the RDIG (Russian Data Intensive Grid) community
created the Russian Tier2 Cluster as WLCG regional facilities for storing and analysis of LC data, and generation of
MC data for all four LHC experiments.
  The results obtained have impact on the deployment of modern information technologies in Russia, on the integration
of Russian science in the European information society, and the creation of grid computing infrastructure in Russia.
Main impact is on the application of modern computing technology in the High Energy Physics, where the Tier2
cluster created as Russian regional center for LCG infrastructure is playing role of successful example for similar
scientific projecvt in other fields, e.g. for creating grid for fusion research (ITER project), in geophysics and biology.
  This project assumes the creation of open source software (for LCG/EGEE projects), thus, there is no direct
economic value or commercial exploitation, as well no patents are planning. However, we hope that such an impact
would be possible in future activity in different application areas by use of our results.
   The research in the framework of this project was based, in particular on the continuation of the collaboration with
colleagues in Europe and other countries within the projects LCG and EGEE-II, including deployment of new
middleware, developing of the grid infrastructure, training etc.
1.4. Summary of Results
 Main results of the project obtained are the following.
  Russia LCG sites (most of them are partners of this INTAS project) are creating the Tier2 cluster with basic Tier2
functionality plus fraction of the Tier1 functionality. This activity was supported partly by this CERN-INTAS grant
and, as result, it has got systematic development according to the LCG plans on creating the global grid infrastructure
of regional centers for LHC data elaborating, storage and for distributed analysis.
  There were elaborated by SINP-MSU team in collaboration with CERN-IT and INFN-Padova teams the modification
of job submission (key component of the LCG2 middleware) allowing transfer of input data to Working Nodes directly
bypassing Resource Broker – this resolves serious bottleneck in existing LCG2 middleware. Then, there was created
tools for monitoring of application jobs, the service useful for users but absent in existing LCG2 middleware. An


                                         CERN-INTAS 03-52-4297 Final Report

                                                       Page 810
                                                                                                        Ref. No 03-52-4297



application monitoring system 'apm' was created by PNPI team allowing grid user to track the execution process of his
job via a web-interface.
   There were created test suites by IHEP team in collaboration with CERN-IT, JINR and PNPI teams for Storage
Resource Manager (SRM) v1.1 and Grid File Access Library (GFAL), fully integrated into the TSTG testing
framework.
  The VMware testbed was created by JINR team in collaboration with FZK team in the Cluster Dubna-Grid for
studying the problem of heterogeneous resources (in many cases operated under Windows OS) potentially available for
HEP applications. This virtualization technology was probed on LHC applications and shown good prospects for
production use in LCG infrastructure. Resulting cluster was used for real work as provision resources for LCG at JINR.
     The Core Infrastructure Center (LCG/EGEE CIC) has been created in SINP-MSU with some functions distributed
at JINR and ITEP – thus distributed model for CIC has been elaborated with full set of basic grid services
implemented. From May 2005 the Russian CIC has started the weekly on-duty operation shifts, rotating with four other
European CICs – CERN, in UK (RAL), France (CC-IN2P3) and Italy (CNAF). The Regional Operational Center
(LCG/EGEE ROC) has been created in IHEP, with distribution of some functions between JINR, ITEP, PNPI and
SINP-MSU, http://grid.sinp.msu.ru/grid/roc/main. The User Call Center has been created by ITEP team in
collaboration with FZK (http://ussup.itep.ru).
  In testing of the distributed model of the CIC and ROC all Russian teams participated, including efforts of
participants from BINP, in particular, taking part in the upgrade of the CERN grid Certification Authority
infrastructure.
    In addition to the comprehensive investigation of the Globus Toolkit 3 executed at the first stage (2004), base
components of following open source toolkits and middleware packages have been studied, installed and tested:
Globus Toolkit 4 (GT4), OMII, and selected gLite components. The work was carried out by SINP-MSU and JINR
teams in close collaboration with the LCG/GTA group at CERN.
   The Russia LCG Web site version 2 was completed, tested and started for the daily providing the LCG users by
JINR team, http://www.egee-rdig.ru.
     The key publications resulting directly from the project are indicated at first ten positions of the above List of
references.
1.5. Role and Impact of INTAS
 Role of INTAS                                 Definitely yes        rather yes        rather not       definitely not
 Would the project have been started                                                                          
 without funding by INTAS?
 Would the project have been carried out                                                   
 without funding from INTAS?


 Main achievement of the project    very important       quite important      less important        not important
 exciting science                                                                    
 new international contacts                                     
 additional prestige for my lab                                 
 additional funds for my lab                                    
 helping scientists in NIS                                      
 other (specify):


 Will the project continue? Directly no.
 Will the co-operation among the project Contractors continue in the future? Yes.

2      MANAGEMENT
2.1. Meetings and Visits
     What co-ordination meetings, exchange visits of scientists, or major field trips took place up to now?
The first coordination meeting of the project was held at JINR (Dubna, Russia) during the Dubna-GRID’2004
conference ("Distributed Computing and Grid Technologies in Science and Education" International Conference,
2004). The meeting was attended by the representative of all the participating Russia teams.


                                           CERN-INTAS 03-52-4297 Final Report

                                                        Page 910
                                                                                                      Ref. No 03-52-4297



 There were three co-ordination meetings at CERN. The preparation meeting was held on 19 March 2004,
http://agenda.cern.ch/fullAgenda.php?ida=a041430. Then two co-ordination meetings were held on 20
September    2004   (http://agenda.cern.ch/fullAgenda.php?ida=a044128)     and     14    March    2005
(http://agenda.cern.ch/fullAgenda.php?ida=a051129). On these meetings representatives of Russian teams
have given talks on the current status of the work, some of them by use of phone conferencing.
     On 25 June 2006 there was held the final project meeting in JINR (Dubna, Russia) attended by all representatives
of the Russia sites and western partners, J. Knobloch, L.Robertson and Z. Sekera from CERN-IT. The Final report was
discussed on this meeting.
     The INTAS travel money have been used basically for covering travel expenses (mostly it was purchasing the
tickets cost) - there were 47 visits to CERN of 15 project participants, in particular for participation in the co-
ordinating meetings. These visits werfe paied partly by tgfhis CERN-INTAS project budget (basically airtickets), and
partly by CERN-IT or home institutes of the russian participants (as in-kind contribution to this INTAS project).
     Four participants (three from JINR, V.Korenkov, V.Mitsyn, E.Tikhonenko, and one from IHEP, E.Slabospitskaya)
participated in the CHEP04 conference in Interlaken (Switzerland, September 2004) with presentation of the results
obtained and for discussions with European colleagues.
     Then, it was visit by V.Korenkov (JINR) to FZK colleagues in November 2004 and in October 2005 for
discussion of the joint work. There was a participation of A.Kryukov (SINP-MSU) in the ACAT2005 Workshop
(Berlin, May 2005) for presentation of the results obtained.
     There were numerous visits of Russian partners each other.
        Summarize the meetings and visits in the table below:
              Visits                            Number of scientists                  Number of person days
          West ==> East                                 3                                      10
          East ==> West                                54                                     652
          West ==> West
           East ==> East                                 35                                     122


2.2. Collaboration
In your opinion how intense was the collaboration among the different Contractors up to now?
   Intensity of Collaboration          high            rather high          rather low                 low
        West <=> East                                       X
        West <=> West                                       X
        East <=> East                   X
In this project we co-operated with the RDIG (Russian Data Intensive Grid) Initiative which drives the activity on
creating the Grid infrastructure for intensive scientific data operations in Russia (considered now as start of National
Grid Initiative) and the EGEE-II international project aiming to develop grid infrastructure for scientific applications,
first of all for LHC experiments.
2.3. Time Schedule
     Up to now, has the time planning been in accordance with the Work Programme?                    Yes
     Do you foresee any deviations from the Work Programme for future? No
2.4. Problems Encountered
     Up to now, did you encounter any major problems ? No
     Summarise your experience in the table below:
 Problems encountered                 Major              Minor                 None              not applicable
 Co-operation of team Members                             X
 Transfer of funds                                        X
 Telecommunication                                                               X
 Transfer of goods                                                               X
 Other


2.5. Actions Required


                                         CERN-INTAS 03-52-4297 Final Report

                                                      Page 1010
                                                                                 Ref. No 03-52-4297



At present, do you have or do you foresee any problem which needs action from INTAS? We had no any
problems which need action from INTAS.




Name of the Co-ordinator: Jürgen Knobloch, CERN IT

Original signature of the Co-ordinator:

Date:




                                     CERN-INTAS 03-52-4297 Final Report

                                                Page 1110

				
DOCUMENT INFO