The load-balancing and server-consolidation policies extensively used by previous virtual infrastructure management tools for cloud-based resource pools, make VM deployment decisions based solely on the demands from the physical resource layer, neglecting the impact of the complementary effects among virtual peers within a distributed service system. On the contrary, this paper provides an efficient deployment algorithm for deploying a P2P network onto a group of physical servers, where each peer encapsulated in the form of VMs, with the purpose of eliminating undesirable side-effects of invalidated replication and snow crash as a result of the recursive replication relations within the peer group. Two improved algorithms, combining both demands from the physical layer (such as load balancing and server consolidation)as well as the P2P service layer (such as peer replication relations), are also provided. Preliminary analysis shows that the proposed algorithms are both correct and cost-fficient in computing proper deployment schemes to avoid invalidated replications while minimizing the likelihood of catastrophic snow crashes triggered by a single physical failure.
[1]
Hui Wang,et al.
Multi-Tiered On-Demand Resource Scheduling for VM-Based Data Center
,
2009,
2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid.
[2]
R. Faure,et al.
Introduction to operations research
,
1968
.
[3]
David R. Karger,et al.
Chord: A scalable peer-to-peer lookup service for internet applications
,
2001,
SIGCOMM '01.
[4]
Jian Chen,et al.
SandStone: A DHT Based Carrier Grade Distributed Storage System
,
2009,
2009 International Conference on Parallel Processing.
[5]
Borja Sotomayor,et al.
Virtual Infrastructure Management in Private and Hybrid Clouds
,
2009,
IEEE Internet Computing.
[6]
Gonzalo Camarillo,et al.
The 3G IP Multimedia Subsystem : Merging the Internet and the Cellular Worlds
,
2004
.
[7]
Arun Venkataramani,et al.
Black-box and Gray-box Strategies for Virtual Machine Migration
,
2007,
NSDI.