Output details
11 - Computer Science and Informatics
Imperial College London
Output title
Camdoop: Exploiting In-network Aggregation for Big Data Applications
Type
E - Conference contribution
DOI
-
Name of conference/published proceedings
USENIX Symposium on Networked Systems Design and Implementation NSDI 12
Volume number
-
Issue number
-
First page of article
29
ISSN of proceedings
-
Year of publication
2012
Number of additional authors
3
Additional information
<06>This paper introduces a novel approach that significantly improves the performance of large-scale data analytics jobs by performing aggregation within the network as opposed to end-hosts only. By exploiting application knowledge, it represents a net departure from common practice that ships bits across the network. This enables up to two orders of magnitude better performance than traditional MapReduce cluster. The insights from this work have been instrumental to obtain two recent grants by EPSRC (NaaS, EP/K032968/1: £666K) and EU (HARNESS, £447K) as well as a PhD fellowship from Google. Usenix acceptance: 17.7%/169
Interdisciplinary
-
Cross-referral requested
-
Research group
B - Distributed Software Engineering
Citation count
-
Proposed double-weighted
No
Double-weighted statement
-
Reserve for a double-weighted output
No
Non-English
No
English abstract
-
