BCube

Abstract

This paper presents BCube, a new network architecture specifically designed for shipping-container based, modular data centers. At the core of the BCube architecture is its server-centric network structure, where servers with multiple network ports connect to multiple layers of COTS (commodity off-the-shelf) mini-switches. Servers act as not only end hosts, but also relay nodes for each other. BCube supports various bandwidth-intensive applications by speedingup one-to-one, one-to-several, and one-to-all traffic patterns, and by providing high network capacity for all-to-all traffic.BCube exhibits graceful performance degradation as the server and/or switch failure rate increases. This property is of special importance for shipping-container data centers, since once the container is sealed and operational, it becomes very difficult to repair or replace its components.Our implementation experiences show that BCube can be seamlessly integrated with the TCP/IP protocol stack and BCube packet forwarding can be efficiently implemented in both hardware and software. Experiments in our testbed demonstrate that BCube is fault tolerant and load balancing and it significantly accelerates representative bandwidthintensive applications.

Interconnection Structures, Management and Routing Challenges in Cloud-Service Data Center Networks: A Survey

“…Designed for shipping containers and modular datacenter networks, the BCube structure is proposed in [32] as a server-centric network interconnection structure that employs multiple commodity switches in a hierarchal-style to interconnect large numbers of multi-port servers.…”

“…Moreover, due to its low diameter, BCube provides high network capacity for all-to-all traffic [32]. Compared to the fat-tree and DCell structures, the authors of [32] claimed that BCube is better than these two structures as it does not have performance bottlenecks, and provides for larger network capacities with direct one-to-x support without minimal upgrade requirements. On the contrary, such direct connection between the network switches and servers imposes high wiring expenses.…”

“…In the case of server or switch failures, such protocol allows for graceful degradation in the bandwidth capacities of the network. The authors of [32] proposed a new BCube routing protocol suite. In their work they claimed to provide a fast packet forwarding protocol that can decide the next hop of any received packet through one table look-up process.…”

“…Consequently, the servers play a double role, where in addition of being end-hosts, they are relay nodes for other communication paths in the network. FiConn [28], DCell [29], and BCube [32] are all structures that fall into this category.…”

Simple, efficient location‐based routing for data center network using IP address hierarchy

“…These designs are termed as server‐centric designs because the routing responsibility lies with the servers rather than intermediate switches. DCell , BCube , MCube , and MDCube are examples for such design. In the second alternative, same features—baseline topology and growth model—help to determine the relative location of a server in a given topology and the obtained location information is used for data transfer.…”