Torrent base of software distribution by ALICE at RDIG

V.V. Kotlyar (IHEP, Protvino), E.A. Ryabinkin (RRC-KI, Moscow), I.A. Tkachenko (RRC-KI,

Moscow),G.S. Shabratova (LHEP JINR, Dubna),

A.K. Zarochentsev (St. Petersburg State University, St. Petersburg)

5th International Conference GRID‘2012"Distributed Computing and Grid-technologies in Science and Education“,

16-21 July, 2012 , Dubna, Russia

AliEn & ALICE Grid

• AliEn: ‘Alice Environment’ – Central Services at CERN

• Task Queue, Job & File Catalog• Software management (build & deployment)

– VO box site-specific operations

• Job Agent submission• Software deployment• Resource monitoring

– Grid Monitoring with MonaALisa

– Data Management

• AliEn File Catalog• Grid-Enabled XRootD SEs

This talk

Types of Software Deployment

Traditional AliTorrent

CentralManagement:

-Additional AliEn torrent store -Catalogue, seeder & tracker

CentralManagement:

-Routine software builds-Catalogue& store in AliEn

Site SW operation:

-Job request SW from VO box service -VO box PackMan service pulls SW-SW deployed on shred area-WNs read SW from shared area

Site SW operation:-VO box is not involved-Jobs pull SW from:• alitorrent.cern.ch speeder• local peers• other sides as available

-though typically behind a FWShared SW area:

-Resource bottleneck-Single point of failure-Can require active repairs per site

Resolves:

-Bottleneck & single point failures-Site level maintenance of shared area

Basic Torrent details

File chunks contain hashes of original file to Provide Data integrity

Tracker: map of seeders:files

Seeders: have & serve file

Leeches: pull & serve file chunks

AliTorrent Details• Torrent Features

– Distributed Hash Tables

• Decentralized seeder lookup – seeders are trackers

– Peer Exchange

• Local peer information is propagated by seeders

– Local Peer Discovery

• Multicast to discover peers on same network

• ALICE/AliEn Features– Total software download is 300-- 400MB‐– Enabled per site (VO box) with AliEn LDAP flag

name=<CN_NAME>,ou=CE,ou=Services,ou=<SITE>,ou=Sites,o=alice,dc=cern,dc=ch installMethod=Torrent

AliTorrent as an opportunity

• AliTorrent use in AliEn– Reduces problems associated with SW deployment– Simplifies site operations by removing a VO Box service

*Does not eliminate VO box model from ALICE Grid* Does eliminate site-specific VO box requirement

• Elimination of site-- specific VO box allows ‐for remote use of other Grid resources: exmp-> OSG

AliTorrent

• ALICE torrent-based software distribution links: (see at http://alitorrent.cern.ch/)– Available software packages – Seeder statistics – Seeder network traffic monitoring – Stand-alone torrent clients

Towards production operations• Site policy caution:

– limits can be broad– has hampered wider

use• ALICE Experience:

– initial distrust turns into recognition of approach’s merit: Stable running at CERN, 2 T1s and a few T2s (including RRC-KI and IHEP)

– simplifies site operations

BitTorrent

Summary• Software deployment on shared area– Bottleneck & site-level single point failure– site-level SW corruption requires admin intervention

• Torrent model AliTorrent– Removes bottleneck & site-level single point of failure– Eliminates a site service & reduces site management– Performance capabilities meets typical ALICE workflow & site requirements Eliminates requirement for site specific VO box‐

• AliTorrent is a site-friendly tool for opportunistic (or general) use– don’t ask the site to “do” something -> install or manage a service– ask the site to “not do” something -> Block Torrent use.

Talk has been prepared on the base:- Report of C. Grigoras at ALICE T1/T2 workshop, Karlsruhe, Germany, January, 2012- Report of R.J. Porter et al at CHEP 2912, New York, USA, May, 2012