TODS: cluster object storage platform designed for scalable services

Authors:
Feng Zhou;Chao Jin;Yinghui Wu;Weimin Zheng
Affiliations:
Computer Science Division, University of California Berkeley, CA;Department of Computer Science and Technology, Tsinghua University, Beijing, China;Department of Computer Science and Technology, Tsinghua University, Beijing, China;Department of Computer Science and Technology, Tsinghua University, Beijing, China
Venue:
Future Generation Computer Systems - Special issue: Advanced services for clusters and internet computing
Year:
2004

Citing 4
Cited 1

Shoring up persistent applications

SIGMOD '94 Proceedings of the 1994 ACM SIGMOD international conference on Management of data
Cluster-based scalable network services

Proceedings of the sixteenth ACM symposium on Operating systems principles
Manageability, availability and performance in Porcupine: a highly scalable, cluster-based mail service

Proceedings of the seventeenth ACM symposium on Operating systems principles
Scalable, distributed data structures for internet service construction

OSDI'00 Proceedings of the 4th conference on Symposium on Operating System Design & Implementation - Volume 4

Relaxation of ACID properties in AuTrA, The adaptive user-defined transaction relaxing approach

Future Generation Computer Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we present the design and implementation of Tsinghua Object Data Store (TODS), a cluster object storage system to support the building of scalable Internet services. TODS provides a unified, transparent and object-oriented view of the storage devices of the whole cluster, which greatly simplifies cluster service development. In the meantime, it is designed to be scalable and efficient. Services built on it can simply inherit these properties in a lot of cases. TODS supports ACID transactions, which facilitates the building of complex transactional services. TODS abstracts away from service logic most complexities of data management, which have often become major obstacles in developing high quality Internet services. The design principles, architecture and implementation of TODS are discussed. In our performance experiments, the system scales smoothly to a 36-node server cluster and achieves 11,160 in-memory reads per second and 396 transactions per second. We also demonstrate that the programming interface is significantly easier to use than that of RDBMS with a head-to-head comparative experiment.