SIGCOMM的一些总结

总的来说，SIGCOMM从08年到12年的论文都比较偏向于Datacenter网络，包括数据中心网络结构、编址、路由、数据中心间大规模数据传递、流管理、网络带宽管理等

 

FairCloud: Sharing The Network In Cloud Computing

Lucian Popa (HP Labs), Gautam Kumar (UC Berkeley), Mosharaf Chowdhury (UC Berkeley), Arvind Krishnamurthy (Univ. of Washington), Sylvia Ratnasamy (UC Berkeley), Ion Stoica (UC Berkeley)

与CPU和存储空间不同，云基础设施提供商很少提供带宽保证机制，因为CPU和内存的竞争域是同一物理机器上的所有虚拟机，而共享同一链路上所有物理机器都存在带宽竞争，导致其分配与保障机制比较复杂。目前基本的带宽分配目标有：min-guarantee、high-utilization、payment-relatived，这三种方式之间存在一定的矛盾关系，而不同类型的服务有不同的带宽需求，例如MapReduce只有在任务分配和汇总的时候有带宽需求，而其他计算阶段无带宽需求。针对该问题，设计了一系列资源分配方式在各种目标间tradeoff。

The Only Constant is Change: Incorporating Time-Varying Network Reservations in Data Centers

Di Xie (Purdue University), Ning Ding (Purdue University), Y. Charlie Hu (Purdue University), Ramana Kompella (Purdue University)

不同的服务有不同的带宽需求特征，例如：constant peak、all peak、fix-width peaks，而固定带宽的分配方式会导致带宽资源的浪费，针对该问题设计了带宽资源预留策略，提高带宽利用率

 

It’s Not Easy Being Green

Peter Xiang Gao (University of Waterloo), Andrew R Curtis (University of Waterloo), Bernard Wong (University of Waterloo), S. Keshav (University of Waterloo)

DONAR: Decentralized Server Selection for Cloud Services

Patrick Wendell (Princeton University), Joe Wenjie Jiang (Princeton University), Michael J. Freedman (Princeton University), Jennifer Rexford (Princeton University

地理分布服务的请求调度

Cloudward Bound: Planning for Beneficial Migration of Enterprise Applications to the Cloud

Mohammad Hajjat (Purdue University), Xin Sun (Purdue University), Yu-Wei Eric Sung (Purdue University), David Maltz (Microsoft Research), Sanjay Rao (Purdue University), Kunwadee Sripanidkulchai (IBM Research), Mohit