Datacast: A Scalable and Efficient Group Data Delivery Service for Data Centers

  • Chuanxiong Guo

MSR-TR-2011-76 |

Reliable group data delivery (RGDD) is a pervasive traffic pattern in data centers. In an RGDD group, a sender needs to reliably deliver a copy of data to all the receivers. Existing solutions either do not scale due to the large number of RGDD groups (e.g., IP multicast) or cannot efficiently use network bandwidth (e.g., end-host overlays).

We design Datacast for RGDD. Datacast turns group state management to in-network packet caching by leveraging the content centric networking (CCN) concept. It uses centrally calculated multiple edge-disjoint trees for efficient and fast data delivery. By introducing a robust rate-based congestion control, which interprets duplicated CCN interest packets as congestion signal, Datacast adapts well to varying network conditions.

We have implemented Datacast using the ServerSwitch platform. Our implementation shows that all the features needed in Datacast, including flexible packet filtering, source routing, and in-network packet caching, can be implemented using commodity devices. Our implementation can offload packet forwarding to switching chip for 50% nodes in a typical group with 280 members and over 500 intermediate nodes. Our testbed experiments demonstrate that Datacast achieves 1.74Gb/s goodput using two 1GbE Steiner trees, it uses network bandwidth efficiently, adapts to changing traffic conditions and reacts to network failures quickly.