Software defined network based adaptive routing for data replication in Data Centers

Abstract

Data availability is a major challenge faced by today's Data Centers where a large number of high performance servers are organized into racks interconnected by a switching network. Data replication is an effective approach for data protection as redundancy ensures at least one copy of data is available in the event of failures. To achieve low latency and high throughput for data replication operations, the traffic load must be spread out evenly across various paths in a Data Center network to minimize congestion. Recently, software defined networking (SDN) has become an attractive solution which enables us to control traffic forwarding in a desired way to achieve our goal. We develop an SDN-based framework for effective data replication operations. We develop an adaptive routing scheme which routes the flows based on the current network state; specifically it chooses routes based on the current load on the paths. Our scheme monitors link-and path-loads and assigns traffic flows to appropriate paths in an efficient way so as to achieve high throughput. We develop a prototype using OpenFlow-enabled switches and carry out experiments to demonstrate the effectiveness of our proposed scheme.

High-availability clusters: A taxonomy, survey, and future directions

“…From a transaction viewpoint, synchronous replication can support all the ACID properties. Therefore, no data loss is usually associated with it [151,97]. However, synchronous replication is challenging with extended distances.…”