Michael Alger http:www.cs.ttu.edualgerm

1
Service-Oriented Programming with Federating
JobbersMaster Thesis Defense, Computer Science -
TTU Oct. 26, 2006

• Michael Algerhttp//www.cs.ttu.edu/algerm
• algerm_at_cs.ttu.edu

2
Speakers Qualifications

• Michael Alger is a computer science graduate
  student at TTU
• Received BS in Computer Science in Spring 2004 at
  TTU
• Interests are in Distributed Computing,
  JavaSpaces, Suns Jini Technology,
  Service-Oriented Architecture, SORCER
• Joined SORCER lab in Spring 2005
• Internships
• Summer Student Research, 2006 Air Force
  Research Laboratory, Wright-Patterson AFB
• Summer Analyst, 2005 Goldman Sachs, NYC

Beginning
3
Presentation Agenda

• Problem Statement
• Hypothesis
• Approach
• Solution FALCON Framework
• Validation Case
• Results
• Benefits
• Conclusion
• Future Work

Beginning
4
Problem Statement

• SORCER Service-Oriented Programming Limitations
• Exertions algorithmic logic does not support
• Iteration
• Branching
• All inner Jobs are coordinated by the same Jobber
  recursively, which may cause a bottleneck

5
Service-Oriented Programming

• All SORCER Provider have service(Exertion) method
• Exertions are the Service-Oriented messages
• Exertion (Method, Context)
• Method (interface, selector)
• Elementary exertions are called ServiceTasks
• Compound exertions are called ServiceJobs
• Job is a composition of Jobs and Tasks
• Jobbers are responsible for handling Jobs

6
Job Execution
Federation
T3
J2
T4
J1
T1
Concatenation
T2
Composition
7
Hypothesis
Algorithmic logic can increase the expressive
power of the Service-Oriented programs in the
SORCER metacomputing environment.
Beginning
8
Approach

• Literature Review SOA and SORCER
• Top-down or Bottom-up approach?
• Design Phase
• SGrid Model
• FALCON Model
• Implementation Phase
• Extend Job runtime execution
• FALCON Framework
• Validation Phase
• Simple differentiation/integration provider
  iteration task
• Conditional job

9
Proposed Solution

• Federated Algorithmic Metacomputing (FALCON)
  framework to support full algorithmic logic
• Iteration
• Branching
• Enable implicit (via Exertion Space) coordination
  for Jobbers and Service Requestors
• Enable explicit (via Catalog) coordination for
  Jobbers

10
FALCON Framework
11
Metacomputer

• Conventional Procedural Programming
• initializeCondition()
• if (ifCondition)
• while(whileCondition)
• derivativeEval f(x)
• else
• while(whileCondition)
• integralEval f(x)

Derivative-Eval
Jobber
Integral-Eval
Initialize-Cond
SO Programming Job job Task initializeCond Task
integralEval Task derivativeEval WhileExertion
whileDer(whileCondition, derivativeEval) WhileExer
tion whileInt(whileCondtion, integralEval) IfExer
tion ifTask(ifCondition, whileDer,
whileInt) job.addExertion(initializeCond) job.add
Exertion(ifTask)
initializeCond
derivativeEval
whileDer
job
then
loop
ifTask
whileInt
integralEval
else
loop
12
What does it mean?

• Build a higher expressiveness in SO Programming

Service Provider
Service Requestor
100 request
1 request
Provider Proxy
Loop
13
Job Coordination
Jobber
Jobber
Jobber
Exertion Space
job/(inner job)
task
explicit
Implicit
ServiceProvider
composite job
ServiceRequestor
Jobber
delegate task
ServiceProvider
ServiceProvider
14
Validation Case 1

• Single variable differentiation or integration

while task
Service Requestor
Derivative-Evaluator
Function-Evaluator
Loop
Condition (deltaResult gt 0.00001 Iteration lt
1000)
15
Validation Case 2

• Conditional Job using previous case

Initialize-Condition
Function-Evaluator
set
then
Service Requestor
Jobber
Derivative-Evaluator
If task
job
loop while task
else
Integral-Evaluator
loop while task
16
Results

• Literature review on SOA and SO programming
• Algorithmic model for SO computing
• UML Design and implementation of the FALCON model
• FALCON deployment in SORCER metacomputing
  environment
• Validation Use Cases

17
Benefits

• Increased expressive power of SO programming
• Iteration
• Branching
• Saves network bandwidth by eliminating redundant
  service request
• Improved performance and scalability by utilizing
  other idle Jobbers
• Exertion Space
• Direct access via Catalog

18
Conclusion

• Scope of the work is more than expected
• Completed the necessary implementation to support
  iteration and branching
• Additional semester
• Model works with the validation case
• Service-oriented base metacomputer
• Very high learning curve from the start
• Exposure

19
Future Work

• Implementation of the full FALCON language
• FALCON language parser for creating Jobs/Tasks
  from an external source
• Update SORCER GUI web portal with new FALCON
  Exertions
• S2S Job/Task load balancing (Preemptive/Non-preemp
  tive scheduling)

20
Summary

• Problem Statement
• Hypothesis
• Approach
• Solution FALCON Framework
• Validation Case
• Results
• Benefits
• Conclusion
• Future Work

End
21
Demo
22
QA