A Remote Computing Broker in .NET

1
A Remote Computing Broker in .NET

• Sendhil Annamalai
• Masters Report, Final Defense
• Under the Guidance of Dr.Daniel Andresen
• Committee Members Dr. Gurdip Singh, Dr. Mitchell
  Neilsen
• Department of Computing and Information Sciences
• Kansas State University

2
Agenda

• Introduction
• Web Services
• Web Service Enhancements
• Implementation
• Technologies Used
• Lessons Learned
• Performance Evaluation
• DEMO
• Conclusion

3
Global Computing

• Cooperative use of geographically distributed
  resources
• Grid computing
• Resource sharing
• Economic Issues
• Security Issues

4
Existing Architectures

• Javelin
• Java-based infrastructure for parallel Internet
  computing
• simple API allowing designers to focus on a
  recursive decomposition/composition of the
  parallelizable part of the computation.
• Globus
• Globus Toolkit - an open source software toolkit
  used for building grids

5
Remote Computing Broker

• Based on .NET
• Remote computation of a .NET binary
• Parallel Internet computation- feasible
• Uses .NET Web Service

6
Web Services

• standardized way of integrating Web-based
  applications
• XML - tag the data
• SOAP - transfer the data
• WSDL - describe services
• UDDI - listing services available
• Advantage of Web Services
• How they work

7
Web References

• The Discovery File
• The WSDL File
• The Reference Map

8
Web Service Enhancements

• Definition
• DIME-Direct Internet Message Encapsulation
• Send and receive SOAP messages with additional
  attachments, like binary files, XML fragments,
  and even other SOAP messages
• Difference between DIME and MIME
• Performance Fast and Efficient
• Simplicity

9
Architecture

• Client - seeks computing resources
• Broker - coordinates between the client and host
  according to the supply and demand for computing
  resources.
• Host - offers computing resources

Host1
Client 1
Broker
Host2
Client N
HostN
10
Detailed Architecture
11
Information Stored

• Host Information
• Bandwidth
• Cost of computation
• Configuration
• Web Reference Url
• Status of Host
• Busy Time
• Client Information
• Input
• Attachment
• Estimated Time
• DeadLine

12
Class Diagram
13
Scheduling Algorithm

• Cost Effective/Time Effective
• Cost List and Configuration List
• Algorithm for selecting cost efficient
• status of first host in the cost list checked
• If it is not busy can assign the job to this host
• If the host is busy and (Busy Time Estimated
  time lt Deadline), assign to this host.
• If the above conditions are not met then the next
  host is considered till the job is scheduled.
• Client can also select the Host

14
Execution of .NET binary
Steps in remote execution of .Net binary
15
Comparison with Javelin
Steps in remote execution of applet
16
Technologies Used

• Visual Studio .NET 1.1
• SOAP
• WSE 1.0
• IIS Server
• Oracle

17
Performance Evaluation

• Aspweb 298 Mhz Dual Pentium 2 Processor, 512
  MB RAM , O/S Windows 2000
• Aspnet 3 Ghz Pentium 4 Processor with
• 2 GB RAM, O/S Windows 2003
• Testing
• Upto 30 files were submitted at the same time.
• API written to test performance
• Testing done with single host and 2 hosts

18
Performance Evaluation
Computation Time for Number of Hosts (aspweb)
19
Performance Evaluation
Computation Time for Number of Hosts-2
20
Performance Evaluation

• Scalability
• Should be able to easily expand to many nodes
• Any number of host can be added
• Fault Tolerance
• Information about hosts stored in database
• Check Host status after estimated time

21

• DEMO

22
Lessons Learned

• Developing Web Applications and writing Web
  Services in .NET
• Programming in different Languages for the same
  project was a new experience.
• Managing and scheduling jobs on many hosts
• Managing Threads and Synchronization primitives
• Using SOAP with attachments.

23
Conclusion

• Remote computing system in .NET is implemented
• Feasible to develop a distributed computing
  system
• System has one broker, can improve the system by
  adding more brokers
• Enhances the performance of the system and
  increases fault tolerance.

24
References

• Bernd O. Christiansen, Peter Cappello, Mihai F.
  Ionescu, Michael O. Neary, Klaus E. Schauser, and
  Daniel Wu, Javelin Internet-Based Parallel
  Computing Using Java
• Michael O. Neary, Alan Phipps, Steven Richman,
  and Peter Cappello, Javelin 2.0 Java-Based
  Parallel Computing on the Internet
• Using Web Service Enhancements to Send SOAP
  Messages with Attachments, http//msdn.microsoft.
  com/library/default.asp?url/library/en-us/dnwebsr
  v/html/wsedime.asp
• Sending Files, Attachments, and SOAP Messages
  Via Direct Internet Message Encapsulation,
  http//msdn.microsoft.com/msdnmag/issues/02/12/DIM
  E/default.aspx
• R. Hundhausen, Building Web Applications with
  ADO. NET and XML Web Services
• Universal Data Discovery and Interchange,
  www.uddi.org
• Understanding DIME and WS-Attachments,
  http//msdn.microsoft.com/library/default.asp?url
  /library/en-us/dnwebsrv/html/dimewsattch.asp
• Webopedia, www.webopedia.com

25
Acknowledgements

• Dr. Daniel Andresen
• Dr. Mitchell Neilsen
• Dr. Gurdip Singh

26

• Questions?