Elastic data routing in cluster-based deduplication systems
Yufeng Wang, Shaojie Tang, Chiu C. Tan
Abstract
Yufeng Wang, Shaojie Tang, Chiu C. Tan
Abstract
As a space-efficient approach to data archive and backup, data deduplication is becoming increasingly popular in storage systems. However, as the data growing rapidly in data centers, single-node storage node is no longer be able to provide the corresponding throughput and capacities as expected. Building deduplication clusters is considered as a promising strategy to leverage such bottle-neck on single-node system. However, deduplication relies on how much the system knows about information of previous stored data. The single-node system obviously obtains all such information and is able to detect duplicate data there; however storage nodes in cluster-based system cannot know information on other nodes. It is nontrivial to route data intelligently enough so that the system could support deduplication performance comparable to that of a single-node system, while also at a trivial cost. In this paper, we propose an elastic data routing strategy, aiming to achieve deduplication performance comparable to state-of-the-art, while require much less computation resources.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
As a space-efficient approach to data archive and backup, data deduplication is becoming increasingly popular in storage systems. However, as the data growing rapidly in data centers, single-node storage node is no longer be able to provide the corresponding throughput and capacities as expected. Building deduplication clusters is considered as a promising strategy to leverage such bottle-neck on single-node system. However, deduplication relies on how much the system knows about information of previous stored data. The single-node system obviously obtains all such information and is able to detect duplicate data there; however storage nodes in cluster-based system cannot know information on other nodes. It is nontrivial to route data intelligently enough so that the system could support deduplication performance comparable to that of a single-node system, while also at a trivial cost. In this paper, we propose an elastic data routing strategy, aiming to achieve deduplication performance comparable to state-of-the-art, while require much less computation resources.
Key concepts: Data deduplication, Computer science, Backup, Node (physics), Computer network, Throughput, Distributed computing, Computer data storage