TY - GEN
T1 - Wide area placement of data replicas for fast and highly available data access
AU - Ping, Fan
AU - Li, Xiaohu
AU - McConnell, Christopher
AU - Vabbalareddy, Rohini
AU - Hwang, Jeong Hyon
PY - 2011
Y1 - 2011
N2 - Recent years have seen rapid growth of online data storage and computing services at various locations around the world. In wide area applications, data can be replicated at multiple locations to serve users with lower latency and higher availability. This paper presents an approach that achieves both fast and highly available data access through periodic migration of data replicas. Such migration strives to maximize a user-defined objective function that incorporates data access delay and availability into a single utility value. To efficiently estimate data access delay and availability for any feasible replica placement, this approach maintains a small data structure that summarizes recent accesses to data replicas. This paper demonstrates, based on an evaluation study, the effectiveness of the developed technique and concludes with plans for future research.
AB - Recent years have seen rapid growth of online data storage and computing services at various locations around the world. In wide area applications, data can be replicated at multiple locations to serve users with lower latency and higher availability. This paper presents an approach that achieves both fast and highly available data access through periodic migration of data replicas. Such migration strives to maximize a user-defined objective function that incorporates data access delay and availability into a single utility value. To efficiently estimate data access delay and availability for any feasible replica placement, this approach maintains a small data structure that summarizes recent accesses to data replicas. This paper demonstrates, based on an evaluation study, the effectiveness of the developed technique and concludes with plans for future research.
KW - Availability
KW - Performance
KW - Replica placement
KW - Replication
UR - https://www.scopus.com/pages/publications/79960174019
U2 - 10.1145/1996014.1996016
DO - 10.1145/1996014.1996016
M3 - Conference contribution
SN - 9781450307048
T3 - DIDC'11 - Proceedings of the 4th International Workshop on Data-Intensive Distributed Computing
SP - 1
EP - 8
BT - DIDC'11 - Proceedings of the 4th International Workshop on Data-Intensive Distributed Computing
T2 - 4th International Workshop on Data-Intensive Distributed Computing, DIDC 2011, held in conjunction with the ACM International Conference on High-Performance Parallel and Distributed Computing, HPDC 2011
Y2 - 8 June 2011 through 8 June 2011
ER -