Second NOW/Cluster Workshop: Building Systems of Systems

Purpose

To examine the progress made since the first NOW/Cluster workshop in October 1994 (ASPLOS VI), assess the current state of the art, and identify the open research challenges in building systems of systems.

Format

The workshop will run for a full day. The format of the workshop is intended to generate a broad-based exchange on areas of current work and directions for solving open problems. It will open with an invited presentation on exciting applications of cluster technology and on emerging cluster interconnect technologies. To achieve a broad based interchange there will be two long poster sessions, and posters will be on display throughout the day. In several key areas, we will have a series of focused discussions led by leading researchers in the field. The purpose of these focused discussions it to provide in-depth coverage of a cluster research topic and a dicussion with deep insights into the key problems and emerging solutions in an area. A closing panel and discussion forum will examine progress and future directions from academic and industry perspectives.

Keynote Presentations

• Making NOW Real: the Inktomi Experience, Eric Brewer, University of California at Berkeley

Focused Discussions

• Emerging Commercial Cluster Interconnects
• Clusters as Storage Servers
• Scheduling Perspectives
• Communications Layers "Shootout"
• Critical Challenges and Future Directions

Preliminary Schedule

Posters

• AM-II: Software and Hardware Support for Virtualized Networks
• Distributed Shared Memory: Is False Sharing a Problem?
• System Administration Solutions
• StarT-Voyager: Message Passing Mechanisms
• StarT-Voyager: Coherent Shared Memory Mechanisms
• Interaction Objects: Parallel OS Support for Exposing StarT-Voyager Communication Resources
• StarT-Jr: A Parallel System from Commodity Technology
• SPADE-2: A Scalable Distributed Shared Memory Parallel Architecture
• Illinois Fast Messages, Version 2.0
• Delivering Communication Performance to the Application: High Level API's atop Illinois Fast Messages
• How to Scale a Transactional Storage System in a Cluster
• Global Energy Simulation on Distributed Memory Systems
• Coupling U-Net with Virtual Memory Management
• Tertiary Disk: A Large Scale Distributed Storage System
• FM-DCS: An Implementation of Dynamic Coscheduling on a Network of Workstations
• Using SCI to Build Shared Memory Clusters
• Implementing High-Level Languages on Fast Networks
• Cilk and Cilk-NOW: Adaptive and Reliable Parallel Computing on Networks of Workstations
• Extending Operating Systems at the User-Level: How to Build Your Own Ufo
• High-Performance Local-Area Communication With Fast Sockets
• Efficient Message Passing in the Avalanche Multiprocessor
• Integrating Tasking and Active Message Style Communication in Gardens
• The Performance of Simple Global Garbage Collector on Large-Scale Distributed Memory Machines
• Quantifying the Effects of Network Performance on End Applications Performance
• The ParaStation Project: Using Workstations as Building Blocks for Parallel Computing
• Illinois Fast Messages on 1Gbps Fibre Channel
• Simplifying Decentralized File Systems Using a Coherent Shared Logical Disk
• Workstation Cluster Software Environment -- MPC++ Approach
• Reducing Waiting Costs in User-Level Communication
• Design and Implementation of Virtual Memory-Mapped Communication on Myrinet
• The Beowulf Project

Related pages

In conjunction with the workshop, we are building a list of cluster projects. If you want, you can register your project.

Here is a page with some related work on clusters.

Go back to the ASPLOS-VII home page

Go to Concurrent Systems Architecture Group Home Page
Go to Berkeley NOW Home Page