Unit:
Κατεύθυνση / ειδίκευση Υπολογιστικά Συστήματα: Λογισμικό και Υλικό (ΣΥΣ)Library of the School of Science
Author:
Προβατάρης Κωνσταντίνος
Supervisors info:
Αλέξης Δελής Καθηγητής
Original Title:
Πειραματική ανάλυση διασωληνωμένης Απεικόνισης-Μείωσης
Translated title:
Experimental Analysis on Pipelined MapReduce
Summary:
MapReduce is a programming model for processing large data sets with a
parallel, distributed algorithm on a cluster. This model is composed of two
execution phases: Map and Reduce, which are distinct and Reduce phase can only
initiate after Map phase is complete. Pipelined MapReduce overlaps total
inactivity of Reduce tasks during Map phase, by pipelining data from Map tasks,
as soon as they are produced, to them. This thesis provides theoretical
analysis on the level of improvement that pipelining can bring, and confirms
the approach with experiments on simulation environments.
Keywords:
Mapreduce, Distributed algorithms, Parallel computing, Data pipelining, Inactivity overlap
File:
File access is restricted only to the intranet of UoA.
document.pdf
1 MB
File access is restricted only to the intranet of UoA.
