OGSA Use Cases

1
OGSA Use Cases

• Aug 19, 2005
• Sogang University
• Distributed Computing Communication Laboratory
• Eunseok, Kim

2
Outline

• Introduction
• Use Cases
• Commercial Data Center
• Severe Storm Modeling
• Online Media and Entertainment
• National Fusion Collaboratory
• Service-Based Distributed Query Processing
• Grid Workflow
• Grid Resource Reseller
• Inter Grid
• Interactive Grids
• Grid Lite
• Virtual Organization Grid Portal
• Persistent Archive
• Mutual Authorization
• Resource Usage Service (RUS)
• Conclusion

3
Introduction

• Document
• The use cases
• Definition of OGSA components
• Prioritization of OGSA components
• Scenarios for many use cases

4
Commercial Data Center

• Consolidation of IT resources
• Outsourcing
• Increasing efficiency
• Decreasing management complexity
• Commercial Data Center
• Commercial GRID System
• Providing Grid-based Meta-OS functionalities
• Scenarios
• Multiple in-house systems
• Use extra IT resource
• Ex) CRM
• Limited time commercial campaign
• Disaster recovery
• Global load balancing

5
Commercial Data Center
6
Severe Storm Modeling

• Predict the exact location of severe storms
• Extremely difficult and expensive
• Not accurate
• Predict severe storms with grid infra
• Combining several components
• Data mining agents, sensors etc
• Not yet possible
• Grid infrastructure is not yet in place

7
Online Media and Entertainment

• Facilitating the ability to dynamically manage
  resources
• based on workload demands and current system
  configuration
• Require Standard interfaces
• Allow dynamic and open collaboration

8
Online Media and Entertainment
9
Online Media and Entertainment Example of
Online gaming system
10
National Fusion Collaboratory

• Fusion magnetic researcher community
• Difficulties
• Install and maintain S/W and machine
• Scientific codes are even more difficult and need
  frequent updates
• Solution
• Application Service Model (a.k.a Network Service
  Model)
• S/W and platform
• Provided or contracted by a service provider
• Accessible remotely to clients.
• Requirements
• Qos-based execution during fusion experiment
• Availability contract (reservation)
• Usage policies (enforce and manage reservation)
• Flexible delegation of rights (analyze usage
  pattern)
• Community accreditation

11
Service-Based Distributed Query Processing using
OGSA and OGSA-DAI

• Service-based distributed query processor
• Evaluating queries expressed in a declarative
  language over one or more existing services
• OGSA-DAI, computational service, etc
• Service orchestration such as workflow languages
• Facilitating discovery and on-demand resource
• Can be cast as a service, Grid Distributed Query
  Service (GDQS)

12
Service-Based Distributed Query Processing using
OGSA and OGSA-DAI

• GDQS
• Factory discovery and service instance creation
  phase
• Resource discovery phase
• Find available grid nodes
• GDQS setup phase
• Query (request) submission phase
• Object Query Language (OQL)
• Query Execution and result delivery phase

13
Service-Based Distributed Query Processing using
OGSA and OGSA-DAI

• Scenarios
• The optimizer
• Chooses from multiple providers
• Spawns multiple copies of an operator
• Exploiting parallelism
• Ex) different GO(Gene Ontology), GIMS (Genome
  Information Management System) databases, BLAST
  services, nodes etc

14
Grid Workflow

• Workflow
• A convenient way of making new services
• By connecting existing services
• Workflow definition
• Registered to a workflow engine
• Then can create a new service
• Grid workflow
• service -gt grid service

15
Grid Resource Reseller

• Grid Resource Reseller
• Provides supply chain between the resource owner
  and end users
• Enable to deal with a few large customers
• Enable to concentrate on their core competence
• Must have ability to switch provider

16
Inter-Grid

• Similar to the Commercial Data Center
• Difference
• It considers geographically-distributed company
  with access to the outside world.
• Aerospace and financial industry
• Features
• Plethora of applications that are not really
  Grid-enabled and difficult to change
• Mixed Grid and non-Grid data centers
• Grid across multiple companies
• Interoperability with Web Services
• Combination of compute/data/service working as a
  single whole
• Scenarios
• Compute/Data/Service Grid

17
Interactive Grids

• Similar to batch processing
• Features
• User interface should be able to interact and
  synchronize seamlessly with processes
• Schedule and perform work based on an automated
  schedule
• Job is being controlled dynamically by an
  external agent
• Computational steering
• User can steer computations and resource needs
  interactively and dynamically

18
Interactive Grids

• Scenarios
• Cases
• UI-based operations controlled by a Grid user
• Pure parallelism and pervasive computing
• User want to monitor and manipulate results in
  real time

19
Grid Lite

• Extends the use of Grids to small devices
• Ex) PDAs, cell phones, firewall etc.
• Grid applications and user can run tasks through
  these Grid-enabled small devices.

20
Virtual Organization Grid Portal

• Grid Portal
• It is assumed that people are members of many VOs
• Provides an end-user view of the collected
  resources available to the members of the VO.
• One-stop shopping
• Usually web-based
• Elements
• Public face
• A set of collaborative tools
• Access to any large data stores
• The ability to make use of any computational
  resources

21
Persistent Archive

• There are many large-data scientific preservation
  environments
• By virtual data grid technology
• Ex) California Digital Library etc
• Preservation Environments handle technology
• Providing appropriate abstraction layers
• Manage mappings between old and new
• Protocols
• Software system
• Hardware system

22
Persistent Archive

• Virtual data Grid
• Derived data product
• Transformative migration of a digital entity to a
  new encoding format
• Managing the completion state
• Describes the result of the application of
  archival processing
• Persistent Archive differs from Virtual data
  Grid
• Adds a transient execution state
• Standard remote data access transparencies
• Logical name space
• Location independent naming convention
• Storage repository abstraction
• Set of operations to be performed on remote
  storage system (ex file system)
• Information repository abstraction
• Set of operations to manage a collection within a
  database
• Access abstraction
• Set of services to be supported by the persistent
  archive

23
Mutual Authorization

• Mutual Authorization
• The most important requirement
• A job submitted to a specific resource on the
  Grid
• Can be authorized by user
• The job may be re-submitted to other resource on
  VO
• Cannot be authorized by user but by VO

24
Resource Usage Service (RUS)

• RUS
• Facilitates the mediation of resource usage
  metrics
• Produced by application, middleware, OS etc
• One of the core services in OGSA
• Exploited by customers
• Measuring resource consumption

25
Conclusion

• All the use cases take advantage of OGSAs
  interoperability.
• Opinion about enhanced MPI
• Framework for workload-based-MPI
• OGSA can provide monitoring capability for user