State and Events for Web Services: A comparison of Five WS-Resource Framework and WS-Notification Implementations

1
State and Events for Web Services A comparison
of Five WS-Resource Framework and WS-Notification
Implementations

• Marty Humphrey, Glenn Wasson, JarekGawor, Joe
  Bester, Sam Lang, Ian Foster, Stephen Pickles,
  Mark McKeown, Keith Jackson, Joshua Boverhof,
  Matt Rodriguez, Sam Meder.
• Presented by Jonatan Alava

2
Background

• What are web services?
• What is WSRF?
• What is WSN?
• And what do they have to do with Grids?

3
Web Services

• Designed to support interoperable
  machine-to-machine interaction over a network.
• Uses a previously described interface (WSDL).
• Communicates using messages via HTTP enclosed in
  a SOAP envelope.
• Allows intercommunication amongst different
  platform and/or programming languages.
• OASIS and the W3C responsible for the
  standardization of web services.
• WS-I established to improve interoperability.

4
WSRF

• Stands for Web Services Resource Framework
• Improves on the concept of Web Services by
  creating a separate view for the resource state.
• Simplifies WSDL and reduces message size and
  complexity (XML gets heavy and complicated fast)

5
WSRF Specification

• WSRF Resource Properties.
• WSRF Resource Lifetime.
• WSRF Base Faults.
• WSRF Service Group.

6
WSN

• Define a set of specifications that standardize
  the way Web services interact.
• Foundations for Event Driven Architectures built
  using Web services.
• "Publish/Subscribe for Web services".

7
WSN Specification

• WS-BaseNotification
• WS-Topics
• WS-BrokeredNotification

8
Web Services and Grids - OGSA

• OGSI problems solved by WSRF

9
OGSA and WSRF

• Open Grid Services Architecture
• The Physiology of the Grid by Ian Foster, Carl
  Kesselman, Jeffrey M. Nick and Steven Tuecke
• Service Orientation and Virtualization.

10
Implementations Compared

• GT4-Java (Argonne)
• GT4-C (Argonne)
• pyGridWare(Lawrence Berkeley Labs)
• WSRFLite(University of Manchester)
• WSRF.NET (University of Virginia)

11
Comparison Parameters

• Transport and SOAP processing.
• Security Issues.
• WS Dispatch and Container.
• Persistence.
• Finding/Discovering WS.
• Lifetime Management.
• Programming and Tooling.
• WS-Notification.

12
Canonical Architecture
13
Key Features
14
Notable fact

• Reuse of existing tools
• GT4-Java Apache Axis, Tomcat
• GT4-C libxml2
• pyGridWare Zolera(ZSI), Twisted
• WSRFLite SOAPLite
• WSRF.NET IIS, ASP.NET, WSE

15
Similarities in Resource Persistence

• By default GT4-Java, GT4-C, pyGridWare and
  WSRFLite use in-memory store.
• Only WSRF.NET uses database store by default.
• All of them offer customization possibilities for
  the persistence store.

16
Differences in Resource Indexing and Retrieval

• GT4-Java and pyGridWare provide ResourceHome
  interface.
• WSRF.NET uses db queries to find resources based
  on resource name or resource state.
• GT4-C and WSRFLite provide own interface and
  default implementations.

17
Differences in Lifetime Management

• GT4-Java and pyGridWare use containers timers to
  periodically purge resources
• GT4-C uses GT4-C common lib timing functions
• WSRFLite varies by resource store
• WSRF.NET uses a windows service to periodically
  perform database deletion queries

18
WSN Issues

• Not all WSN specs implemented by all.
• WSRF.NET only implementation that includes
  complete WSN features.
• GT4-Java and pyGridWare implement
  WS-BaseNotification and WS-Topics.
• GT4-C implements client-side notification only.
• WSRFLite currently does not support WSN.

19
Performance Test

• Systems implemented a Counter Service.
• Single Resource Property counter_val
• GetRPtest ? get counters value.
• SetRPtest ? set counters value.
• Create test ? create new counter resource.
• Destroy test ? destroy resource.
• Notify test ? send state change notifications to
  subscribed clients.

20
Test Scenarios

• 6 test scenarios
• Client and Service on same machine or on separate
  machines with
• No security.
• Transport security.
• Message security.

21
Performance Evaluation (NO SECURITY)
No Security
22
Performance Evaluation (TRANSPORT SECURITY)
Transport Level Security
23
Interoperability

• Many issues in interoperability.
• At the time the paper was published the
  implementation were using different versions on
  WSRF.
• Non trivial problem for the future due to
  problems outside the scope of WSRF.

24
In Closing

• Paper offers a very brief look at WSRF
  implementations in the context of OGSA.
• Not a real comparison of advantages and
  disadvantages.

25
References

• Wikipedia. www.wikipedia.com
• WS-Resource FrameworkGlobus Alliance
  Perspectives a presentation by Ian Foster found
  at www.globus.org
• Web Services Resource Framework (WSRF) Primer
  from OASIS found at docs.oasis-open.org/wsrf/wsrf-
  primer-1.2-primer-cd-01.pdf
• W3C at www.w3c.org
• Web Service Notification (WSN) Specification
  from OASIS found at docs.oasis-open.org/wsn/wsn-ws
  _base_notification-1.3-spec-pr-02.pdf
• State and Events for Web ServicesA Comparison of
  Five WS-Resource Framework and WS-Notification
  Implementations a presentation by Glenn Wasson
  from the 14th IEEE International Symposium on
  High Performance Distributed Computing (HPDC-14)
  found at www.caip.rutgers.edu/hpdc2005/presentatio
  ns/session1-gwasson.pdf