BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Using constraint programing to resolve the multi-source / multi-si
 te data movement paradigm on the Grid
DTSTART;VALUE=DATE-TIME:20081104T153500Z
DTEND;VALUE=DATE-TIME:20081104T160000Z
DTSTAMP;VALUE=DATE-TIME:20130518T221051Z
UID:indico-contribution-167@cern.ch
DESCRIPTION:Speakers: Mr. ZEROLA\, Michal (Nuclear Physics Inst.\, Academy
  of Sciences\, Praha)\nIn order to achieve both fast and coordinated data 
 transfer to collaborative sites as well as to create a distribution of dat
 a over multiple sites\, efficient data movement is one of the most essenti
 al aspects in distributed environment. With such capabilities at hand\, tr
 uly distributed task scheduling with minimal latencies would be reachable 
 by internationally distributed collaborations (such as ones in HENP) seeki
 ng for scavenging or maximizing on geographically spread computational res
 ources. But it is often not all clear (a) how to move data when available 
 from multiple sources or (b) how to move data to multiple compute resource
 s to achieve an optimal usage of available resources.\n\nConstraint progra
 mming (CP) is a technique from artificial intelligence and operations rese
 arch allowing to find solutions in a multi-dimensional space of variables.
  We present a method of creating a CP model consisting of sites\, links an
 d their attributes such as bandwidth for grid network data transfer also c
 onsidering user tasks as part of the objective function for an optimal sol
 ution. We will explore and explain trade-off between schedule generation t
 ime and divergence from the optimal solution and show how to improve and r
 ender viable the solution's finding time by using search tree time limit\,
  approximations\, restrictions such as symmetry breaking or grouping simil
 ar tasks together\, or generating sequence of optimal schedules by splitti
 ng the input problem.\n\n\nResults of data transfer simulation for each ca
 se will also include a well known Peer-2-Peer model\, and time taken to ge
 nerate a schedule as well as time needed for a schedule execution will be 
 compared to a CP optimal solution. We will additionally present a possible
  implementation aimed to bring a distributed datasets (multiple sources) t
 o a given site in a minimal time.\n\nhttp://indico.cern.ch/contributionDis
 play.py?contribId=167&sessionId=29&confId=34666
LOCATION:Ettore Majorana Foundation and Centre for Scientific Culture
URL:http://indico.cern.ch/contributionDisplay.py?contribId=167&sessionId=2
 9&confId=34666
END:VEVENT
END:VCALENDAR