Computational Grids Aaron Witt EECC 756 May 18, 1999
Introduction What are they? Why do we need them? Who will use them?
Evolution ClusteringClustering Heterogeneous Distributed ComputingHeterogeneous Distributed Computing MetacomputingMetacomputing
Grid Characteristics Scaling and selectionScaling and selection HeterogeneityHeterogeneity Unpredictable structureUnpredictable structure Dynamic and unpredictable behaviorDynamic and unpredictable behavior Multiple administrative domainsMultiple administrative domains
Applications Distributed supercomputingDistributed supercomputing High-throughputHigh-throughput On-demandOn-demand CollaborativeCollaborative Data-intensiveData-intensive
The Globus Toolkit Resource location and allocation (GRAM)Resource location and allocation (GRAM) Communications (Nexus)Communications (Nexus) Resource information service (MDS)Resource information service (MDS) Authentication interface (GSI)Authentication interface (GSI) Executable management (GEM)Executable management (GEM) Remote data access (GASS)Remote data access (GASS) Health and status (HBM)Health and status (HBM)
The Globus Toolkit
The GUSTO Testbed Globus Ubiquitous Supercomputing TestbedGlobus Ubiquitous Supercomputing Testbed 47 Sites47 Sites 2.5 Tflop/s2.5 Tflop/s
Future Direction AWARE (Adaptive Wide Area Resource Environment)AWARE (Adaptive Wide Area Resource Environment) Increase test bed sizeIncrease test bed size Quality of ServiceQuality of Service IntegrationIntegration