MCube: A high performance and fault-tolerant network architecture for data centers
Authors:
- 1 June 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 5, V5-423-V5-427
Abstract
A fundamental challenge in data center networking is how to efficiently interconnect a huge amount of servers. The network architecture typically consists of a tree of routing and switching elements with progressively more specialized and expensive equipment moving up the network hierarchy. Unfortunately, even when deploying the highest-end IP switches/routers, resulting topologies remain complicate application design and limit overall system performance. This paper presents MCube, a server-centric network architecture specifically designed for data centers. MCube leverages commodity mini-switches to provide high aggregate bandwidth and can deliver more performance at less cost than available from today's higher-end solutions. MCube is fully backward compatible with Ethernet, IP, and TCP and provides higher network capacity than the traditional tree-based structure for various types of services. Furthermore, MCube is also fault tolerant since it does not have single point of failure and its source routing protocol performs near shortest-path routing even in the presence of severe link or node failures. Experiences show that MCube is fault tolerant and load balancing and it significantly accelerates the bandwidth-intensive applications.Keywords
This publication has 9 references indexed in Scilit:
- VL22009
- BCube2009
- FiConn: Using Backup Port for Server Interconnection in Data Centers2009
- Towards a next generation data center architecture2008
- A scalable, commodity data center network architecture2008
- DcellACM SIGCOMM Computer Communication Review, 2008
- Dryad2007
- Graph-theoretic analysis of structured peer-to-peer systems2003
- The Google file system2003