网络数据信息传输仅使用传统单独信道、集中式存储的单播方式，难以满足海量用户数据放置与存储需求，通过在数据中心引入涵盖多条可靠信道的可靠组、Steiner树算法、RDT(Reliable Data Transfer Protocol)数据传输协议，代替传统的单播信道、Hash分布式键值存储的数据通信方案，可建立起组播拥塞控制的增量部署模型，根据数据中心网络拓扑结构构造子边不相交的数据控制生成树，将数据源数据分发至多个可靠组的网络信道进行传输，使用AIMD(Additive Increase and Multiplicative Decrease)算法做出组播数据传输的拥塞控制，进而提升Datacast数据传输模式的网络带宽利用率。
The research on topology and congestion control of data transmission networks based on data centers
The transmission of network data information only uses traditional single channel and centralized storage unicast methods,which are difficult to meet the needs of massive user data placement and storage,by introducing reliable groups,Steiner tree algorithm,and RDT data transmission protocol covering multiple reliable channels in the data center,replacing traditional data communication schemes such as unicast channels and Hash distributed key value storage,an incremental deployment model for multicast congestion control (Datacast) is established.Based on the topology of the data center network,a data control spanning tree with non overlapping sub edges is constructed to distribute data source data to the network channels of multiple reliable groups for transmission.The AIMD algorithm is used to control multicast data transmission congestion,thereby improving the Datacast performance.The network bandwidth utilization of data transmission mode.
data centernetwork topology structurescalabledata transmission
万方数据
维普
