Archive of UserLand's first discussion group, started October 5, 1998.
Distributed OO Network Application Architecture (was zzzzzz)
Author: Matthew Dornquast Posted: 1/27/1999; 12:37:40 PM Topic: Jini Jini Jini (Zzzzz?) Msg #: 2537 (In response to 2535) Prev/Next: 2536 / 2538
Another product deserves consideration (IMHO) when discussing distributed computing/java. It's Objectspace's voyager. Check it out! <http://www.objectspace.com/developers/voyager/white/index.html>
A few key services/technologies in distributed computing: Location transparency (Code doesn't think of being at machine X, it thinks of running near a service or chunk of data-- wherever that is) Federated directory/naming services Autonomous code(self moving/agent) Pass by ref AND value. (ala corba 3.0)
Other tools you need to make that stuff work: Transaction services (multi-phase commit across multiple cpus) Persistence services (one could argue that a large enough space with distributed cloning could eliminate much of this requirement.
When you combine: object oriented development (in particular providing solutions via component collaboration), distributed computing technologies (listed above), and Java's VM/classloader w/ OO infrastructure you get something that's well. Unique in my development experience.
I can *see* solutions through tightly integrated systems following strict open standards in a hardwired web of applications.
I *dream* about collaborations of loosely coupled objects whose concerns are roles, responsibilities and services.
It's a different way to architect. One we didn't have the luxury of doing before the internet/java. (Okay, some did-- and DID. But they were proprietary systems with limited space to grow)
There is no longer any reason to sacrifice the flexibility of an application's architecture to achieve the required performance goal.
The precondition to such a statement is the application runs in a networked environment with sufficient local resources to establish a distributed architecture on which to work.
We've got enough technology/io/cpu/network cycles to distribute any problem accross N machines.
If we don't, can anyone argue we wont?>>
This page was archived on 6/13/2001; 4:47:37 PM.
© Copyright 1998-2001 UserLand Software, Inc.