Title: JAVELIN
1JAVELIN
Michael D. Myjak Vice President CTO The Virtual
Workshop, Inc. P.O. Box 98 Titusville, FL
32781 mmyjak_at_virtualworkshop.com
Sean T. Sharp Sr. Software Engineer The Virtual
Workshop, Inc. P.O. Box 98 Titusville, FL
32781 ssharp_at_virtualworkshop.com
2JAVELIN
Java Virtual Environment Layered Infrastructure
- Java Real-Time RTI
-
- the Virtual Reality Namespace Service
3JAVELIN VRNS - Overview
- In the Beginning
- Darwinian Simulation
- Theres a New Kid in Town
- The Nuts and Bolts of VRNS
- Future Plans
4In the Beginning.
- Jack A. Thorpes paper entitled
- "Future Views Aircrew Training 1980-2000.
- Advances are on the Horizon
- Not just Simple Improvements
- Bold New Concepts
- Became the Catalyst for Large-Scale Networked
Simulation Systems
5The Cyberspace Battlefield
- The Quest Began in the Late 1980s
- SIMulator NETworking (SIMNET)
- Distributed Interactive Simulation
- Aggregate Level Simulation Protocol
- Highly Latent Architecture
- REALLY Troublesome Interface
6The Cyberspace Battlefield
- The Quest Began in the Late 1980s
- SIMulator NETworking (SIMNET)
- Distributed Interactive Simulation
- Aggregate Level Simulation Protocol
- High Level Architecture
- Run Time Infrastructure
7Simulating Moores Law
- Hardware Performance Doubles
- Semiannually
- Simulation programs continue to evolve in both
complexity and performance - And at a rate much faster then the protocols upon
which they were based. - This eventually leads to a situation that is
untenable protocols that are too rigid for the
task at hand.
8Theres a New Kid in Town!
- A Couple of Concurrent Events
- Object Oriented Programming Constructs (a late
70s concept) - A Stable Internet that Worked
- Flexible, Constructive Simulations
- And of course, that Interactive Stuff based on a
stodgy, rigid, out-dated protocol
(that actually worked)
9Platform Proto Federation
- Early experience foreshadowed many of the events
yet to come
10Exploring the Weaknesses in HLA
- IF we dont learn from PAST mistakes then were
bound to repeat them! - So why does protocol-based Interoperability work
anyway? - Occums Razor?
11A New Opportunity
- National Information Infrastructure (NII)
Initiative - Sub to the Naval Postgraduate School
- Lead by Jaron Lanier
- Attempted to Integrate Human/VR/Simulation
through NII - A New Idea sparked a catalyst...
- A Scaleable VR Namespace Service
12Simple Ideas or Software Alchemy
- Zyda postulated the advantages of a VR Name
Server to Net-VEs - Brutzman was seeking a methodology that would
permit dynamic protocols - Federated Simulations had the FOM
- Hmmm...
- Put all these concepts in a blender...
- - Could we create large, scaleable Networked
Virtual Environments ?
13So what if...
- Could we provide a methodology to store raw
object classes, methods and their attributes
on-line? - Yeah, well thats been done.
- So what if we could also store object instances
and their behaviors on-line? - Thats been done before too
- Now apply a component-based approach
14Software Gestalt
- Give names to these objects, methods, attributes,
instances, protocols, etc. - Store these objects on-line, in a readily
accessible manner (read automated) - Develop a baseline methodology for a scaleable
(i.e., hierarchical) approach - Layout an XML/DTD description of these objects...
and a new methodology is born!
15Dynamic Semantics...
- Dynamically alter the state information and/or
behavior of objects executing in an operational
environment. - Not only could we store and recall
- object/protocol information
- object instances and alter methods
- We managed to demonstrate that the semantics of
object protocol behaviors could be altered
dynamically - While those objects are running
16JAVELIN - Revolutionary New Technology
- We began by integrating our concepts into Javelin
- FOM/RID file was already in XML
- URLs were already being used in Javelin to load
the FOM/RID information apriori - Now we had a mechanism to alter the FOM data
dynamically...
17Flexible FOMs
- The concept of an agile FOM isnt new
- An implementation strategy for dynamic FOMs and
RTIs certainly is - A mechanism which could permit the FOM RTI data
structures to be altered dynamically could
incorporate - Dynamic FOMs and Agile Run Time Infrastructures
- Permit New Federates to join an Executing
Federation - Late Joiners with New Object Classes
- Efficient Departure of Dead Federates Objects
- And automate the garbage collection of RTI state
info. - Ad hoc Federation Deployment
18Just In Time Operations...
- Performance Issues were initially explored
- Used Open LDAP for serving VRNS
- Not the speediest of performers
- 50 msec to update method or class
- Commercial LDAP systems perform better
- Scaleable, but tools to manage the directory
structure need to be developed
- Dynamic Configuration Manglement
19Just In Time Operations...
- Performance Issues were initially explored
- Used Open LDAP for serving VRNS
- Not the speediest of performers
- 50 msec to update method or class
- Commercial LDAP systems perform better
- Scaleable, but tools to manage the directory
structure need to be developed
- Dynamic Configuration Management
20Summary
- A Proof of Principle Demonstration Showed that
Dynamic Semantics Work - Java work-arounds explored demonstrated
- Used Lightweight Directory Application Protocol
(subset of X.500) - Noted that the Directory Structure plays a Direct
Impact on performance - Culling data background management are just as
important as in operational use.
21Plans
- Continue to do research in this area
- Currently making Alterations to Javelin to
support dynamic FOMs and dynamic internal RTI
state - Integrating with ADL LMS concepts
- Seeking funding to support follow-on work to
extend these concept(s)
22Got Questions?
23JAVELIN
Michael D. Myjak Vice President CTO The Virtual
Workshop, Inc. P.O. Box 98 Titusville, FL
32781 mmyjak_at_virtualworkshop.com
Sean T. Sharp Sr. Software Engineer The Virtual
Workshop, Inc. P.O. Box 98 Titusville, FL
32781 ssharp_at_virtualworkshop.com