A comprehensive guide to state-of-the art clusters technology for professionals and graduate-level CS students. Covers hardware as well as software issues, including topics such as: active messages and lightweight protocols, message passing systems. Cluster computers can deliver better performance than traditional mainframes and super computers at a much lower hardware cost.
Key Feature * Includes contributions from more than 100 of the field's leading practitioners and researchers. * Networking, lightweight protocols, I/O, programming tools, process scheduling, load sharing, and much more. * State-of-the-art techniques for enhancing availability and throughput.
Table of Contents Part I: Requirements and General Issues. 1. Cluster Computing at a Glance. 2. Cluster Setup and its Administration. 3. Constructing Scalable Services. 4. Dependable Clustered Computing. 5. Deploying a High Throughput Computing Cluster. 6. Performance Models and Simulation. 7. Metacomputing: Harnessing Informal Supercomputers. 8. Specifying Reasource and Services in Metacomputing Systems.
Part II: Networking, Protocols and I/O. 8. High Speed Networks. 9. Lightweight Messaging Systems. 10. Active Messages. 11. Xpress Transport Protocol. 12. Congestion Management in ATM Clusters. 13. Load Balancing Over Networks 14. Multiple Path Communication. 15. Network RAM. 16. Distributed Shared Memory. 17. Parallel I/O for Clusters: Methodologies and Systems. 18. Software RAID and Parallel Filesyatems.
Part III: Process Scheduling, Load Sharing, and Balancing. 20. Job and Resource Management Systems. 21. Scheduling Parallel Jobs on Clusters. 22. Load Sharing and Fault Tolerance Manager. 23. Parallel Program Scheduling Techniques. 24. Customized Dynamic Load Balancing. 25. Mapping and Scheduling on Heterogeneous Systems.
Part IV: Representative Cluster Systems. 26. Beowulf. 27. RWS PC Cluster II and Score Cluster System Software. 28. COMPaS: A Pentium Pro PC-Based SMP Cluster. 29. The NanOS Cluster Operating System. 30. BSP-Based Adaptive Parallel Processing. 31. Mars: An Adaptive Parallel Programming Environment. 32. The Gardens Approach to Adaptive Parallel Computing. 33. The Par Par System: Software MPP. 34. Pitt Parallel Computer. 35. The RS/6000 SP System: A Scalable Parallel Cluster. 36. A Scalable and Highly Available Clustered Web Server.