Clustered Multicast Source Routing for Large-Scale Cloud Data Centers

Abstract

The multi-tenancy concept in cloud data center (DC) networks paves the way towards advancements and innovation in the underlying infrastructure such as network virtualization. Multicast routing is essential in leveraging multi-tenancy to its full potential. However, traditional IP multicast routing is not suitable for DC networks due to the need to support a massive amount of multicast groups and hosts. State-of-the-art DC multicast routing approaches aim to overcome these scalability issues by, for instance, taking advantage of the symmetry of DC topologies and the programmability of DC switches to compactly encode multicast group information inside packets, thereby reducing the overhead resulting from the need to store the states of flows at the network switches. Although these approaches scale well with the number of multicast groups, they do not perform well with group sizes and, as a result, yield substantial traffic control overhead and network congestion. In this article, we present Bert, a scalable source-initiated DC multicast routing approach that scales well with both the number and size of multicast groups through the clustering of multicast group members where each cluster employs its own forwarding rules. Compared to the stateof-the-art approach, Bert yields much less traffic control overhead by significantly reducing packet header sizes and eliminating switch memory usage across the switches.

Ernie: Scalable Load-Balanced Multicast Source Routing for Cloud Data Centers

“…In these approaches, a route between source and destination(s) is defined as the remainder of the division between a route-small DCs (micro data centers), and do not scale beyond a few tens of switches. Recent source-routed schemes, like Elmo [15] and Bert [16], address both control and data planes scalability limitations by exploiting DC topology symmetry as well as hardware switch reconfigurability. Although, these are shown to scale well with millions of multicast groups, they still present some major issues.…”

“…It behaves poorly under these scenarios by increasing network overhead (i.e., switch memory and packet size overheads). On the other hand, Bert [16] imposes bandwidth and end-host CPU overheads when the number of clusters (packet replications at the source) is large.…”

“…In multicast source-routing approaches, a multicast tree is encoded into the header of each packet, and switches can read this information to make forwarding decisions. Recent source-routed schemes [15], [16] address both control and data planes scalability limitations by exploiting DC topology symmetry as well as hardware switch reconfigurability. Topology symmetry of DC networks (e.g.…”

“…Recently proposed source-routed multicast schemes for cloud DCs, such as Elmo [15] and Bert [16], address the scalability limitations and are shown to scale well with millions of multicast groups. They do so by exploiting both the DC network topology symmetry and hardware switch programmability to efficiently encode multicast routing information inside packets.…”

“…For instance, Elmo [15] incurs extra overhead when the multicast group is large in size or dispersed across the network. Although Bert [16] aims to alleviate network overhead incurred by Elmo, it imposes bandwidth and end-host CPU overheads when the number of clusters (packet replications at the source) is large. Furthermore, these schemes neglected the multicast traffic load balancing and relied on underlying multipathing protocols (e.g., ECMP [17]).…”