SEARCH
ISTC-CC NEWSLETTER
RESEARCH HIGHLIGHTS
Ling Liu's SC13 paper "Large Graph Processing Without the Overhead" featured by HPCwire.
ISTC-CC provides a listing of useful benchmarks for cloud computing.
Another list highlighting Open Source Software Releases.
Second GraphLab workshop should be even bigger than the first! GraphLab is a new programming framework for graph-style data analytics.
ISTC-CC Abstract
Net-Cohort: Detecting and Managing VM Ensembles in
Virtualized Data Centers
Proceedings of 2012 IEEE International Conference on Autonomic Computing (ICAC'12),
Liting Hu1, Karsten Schwan1, Ajay Gulati2, Junjie Zhang1, Chengwei Wang1
1College of Computing,
Georgia Institute of Technology
2Resource Management Team,
VMware, Inc.
Bi-section bandwidth is a critical resource in today's data centers because of the high cost and limited bandwidth of higher-level network switches and routers. This problem is aggravated in virtualized environments where a set of virtual machines, jointly implementing some service, may run across multiple L2 hops. Since data center administrators typically do not have visibility into such sets of communicating VMs, this can cause inter-VM traffic to traverse bottlenecked network paths. To address this problem, we present 'Net-Cohort', which offers lightweight system-level techniques to (1) discover VM ensembles and (2) collect information about intra-ensemble VM interactions. Net- Cohort can dynamically identify ensembles to manipulate entire services/applications rather than individual VMs, and to support VM placement engines in co-locating communicating VMs in order to reduce the consumption of bi-section bandwidth. An implementation of Net-Cohort on a Xen-based system with 15 hosts and 225 VMs shows that its methods can detect VM ensembles at low cost and with about 90.0% accuracy. Placements based on ensemble information provided by Net- Cohort can result in an up to 385% improvement in application throughput for a RUBiS instance, a 56.4% improvement in application throughput for a Hadoop instance, and a 12.76 times improvement in quality of service for a SIPp instance.
FULL PAPER: pdf
