分享到微博 分享到人人 分享到LinkedIn 分享到Email
MSRA-UCLA IPAM Student Summer School 2010 Project Introduction
作者 微软研究院
2010年3月5日

Research Group: System
Researcher: Lidong Zhou
Project Name: Scheduling and Resource Management for Cloud OS
Project Description:
In this project, we are looking at a specific type of distributed computing infrastructure for cloud computing, namely, a data parallel distributed computing system such as Map/Reduce, Hadoop, SCOPE, and Dryad/DryadLINQ.
Reference:
[1] Jeffrey Dean, Sanjay Ghemawat: MapReduce: Simplified Data Processing on Large Clusters. OSDI 2004: 137-150
[2] Michael Isard, Mihai Budiu, Yuan Yu, Andrew Birrell, Dennis Fetterly: Dryad: distributed data-parallel programs from sequential building blocks. EuroSys 2007: 59-72
[3] Michael Isard, Vijayan Prabhakaran, Jon Currey, Udi Wieder, Kunal Talwar, Andrew Goldberg: Quincy: fair scheduling for distributed computing clusters. SOSP 2009: 261-276


Research Group: Wireless and Networking
Researcher: Chuanxiong Guo
Project Name: Bandwidth Guarantee for Virtual Data Centers in the Hose Model
Project Description:
In this project, we will study the following two problems for bandwidth allocation in the Hose model.
1. Given a set of VMs been allocated to a set of physical servers and each VM with an egress/ingress bandwidth requirement, suppose we know the traffic matrix (e.g., by using certain measurement method), how can we perform flow scheduling so that the requirements of the user can be satisfied.
2. Given a set of VMs been allocated to a set of physical servers and each VM with an egress/ingress bandwidth requirement, suppose we do not know the traffic matrix, how to reserve bandwidth in the network, so that we can accept all the feasible traffic matrices.
Reference:
[1] Amit Kumar et al. Algorithms for Provisioning Virtual Private Networks in the Hose Model, in ACM SIGCOMM 2001.
[2] A. Greenberg et al. VL2: A Scalable and Flexible Data Center Network. In SIGCOMM, Aug 2009.
[3] C. Guo et al. BCube: A High Performance, Server-centric Network Architecture for Modular Data Centers. In SIGCOMM, Aug 2009.
[4] R. Mysore et al. PortLand: A Scalable Fault-Tolerant Layer 2 Data Center Network Fabric. In SIGCOMM, Aug 2009.


Research Group: Internet Media
Researchers: Feng Wu, Chong Luo
Project name: Understanding the properties of arithmetic graph code from information theory perspective.
Project Description:
In this project, we propose to study the proprieties of arithmetic graph code. To make our analysis simple, we can limit that the number of edges of each right and left nodes has unique value, respectively.
Reference:
[1] T. Richardson, R. Urbanke, “Modern Coding theory,” Cambridges 2008.


Research Group: Data Mining
Researcher: Haixun Wang
Project Name: Probabilistic Matching in Uncertain, Large Graphs
Project Description:
We want to develop a system and method for probabilistic matching of graphs. The information in the graph is uncertain. For example, the existence of an edge, of a label on a node, and of a label on an edge are probabilistic.
Reference:
[1] Managing and Mining Graph Data, Springer, Ed. C. Aggarwal, H. Wang, 2010.
[2] Probabilistic Graph and Hypergraph Matching, R. Zass, CVPR 2008.
[3] Probabilistic graph matching by canonical decomposition, Yaghi, H. Krim, H. North, ICIP 2008.
[4] RDF Semantics, P. Hayes, W3C Recommendation, 2004.
[5] The YAGO-NAGA Approach to Knowledge Discovery, G Kasneci, F Suchanek, M. Ramanath, G Weikum, SIGMOD Record 37:4, 2008