| Weblog: | Apple As Innovator | |
| Subject: | Jini and Rendezvous: Apples and Oranges | |
| Date: | 2003-05-29 14:59:13 | |
| From: | anonymous2 | |
|
Jini and Rendezvous have little in common. They both support service discovery and that is about it. Rendezvous does nothing to solve many of the interesting networking problems when dealing with networked services. Allocation of service resources with leasing, availabilty of servcies with activation, dealing with service events, and distributed transactions are all dealt with in Jini and decidedly not in Rendezvous. With Rendezvous the application or OS must understand the service protocol or SOL. There are *NO* protocols with Jini. At least none you would see in dealing with a Jini service. So after all these years all we get is sevice discovery? Excuse me for being less than impressed.
|
||
Showing messages 1 through 5 of 5.
-
Jini and Rendezvous: Apples and Oranges
2003-05-30 04:59:42 Daniel H. Steinberg |
[Reply | View]
There are two things that Rendezvous got right: (1) They explained their technology in a compelling way to consumers -- here's what it can do for you. Conferences with wireless are changed with Rendezvous. (2) They explained their technology in a compelling way to hardware and software vendors so that we've seen shipping applications within the first six months of Rendezvous' public announcement.
Do Jini and Rendezvous have little in common? I would agree that Rendezvous doesn't solve the leasing and transaction and other problems that you detail above. But one of the biggest barriers to using Jini has been the "out of the box" problem. Getting it up and started.
I would argue that Jini and Rendezvous complement each other nicely. A Rendezvous enabled Jini application could lead to local dynamic grids built around JavaSpaces. Making the Jini lookup service a Rendezvous service could have interesting applications.
-
Jini and Rendezvous: Apples and Oranges
2003-06-01 17:57:09 anonymous2 [Reply | View]
To me as a user of technology rather than someone who understands the guts of things, Rendezvous is adding back some of the best features of Appletalk to the TCP/IP world. It is back to the future. The ability to identify printers and colleagues on the network and interact with them simply is just Apple getting everybody to do what we used to do all the time. It will advance the digital hub idea greatly.
-
Jini and Rendezvous: Apples and Oranges
2003-05-29 15:28:36 Tim O'Reilly |
[Reply | View]
This comment reminds me of the disparaging comments that SGML proponents made about HTML and the WWW. Ditto the folks who were building GUI apps, or beautiful multimedia CD-ROMs. All of them saw how much was lacking in the WWW. But sometimes, less is more. This is a major point of Clayton Christenson's book, The Innovator's Dilemma.
We have to ask ourselves why Jini and Jxta never took off? I think that the answer is that they tried to solve too many problems too early. My guess is that much of their work will be rediscovered. But right now, we just need a low-barrier-to-entry method for getting started in ad-hoc networking, and Rendezvous seems to fill the
-
Jini and Rendezvous: Apples and Oranges
2003-05-29 15:22:39 tychay [Reply | View]
Change a few words around and put it in a different time period (1993) and you could make the same case about HTML & HTTP (web protocols) vs. everything else out at the time.
HTML/HTTP does little to solve many of the interesting documentation problems: A primitive bastardization of SGML designed to be more human-readable than machine read (no XHTML) that mixes logic and content (<strong> and <b>), capable of not even the basic in information design and aesthetics (not even a <table> with single pixel gif, no stylesheets), that is stateless (no cookies), bandwidth inefficient (no consistent connection, TCP only), non-interactive (no java or plugin architecture, just helper apps), insecure (no SSL, passwords passed base64encoded) with a toy protocol that can't even automatically validate link consistency.
To paraphrase: ``When you say that the Web got it right, you must mean that Tim Berners Lee and others just did a better job of putting a lesser technology to more visible and actual use.''
Sometimes, it's all about hitting the "sweet spot". Which, I think, is Tim O'Reilly's point.
Take care,
terry
| Showing messages 1 through 5 of 5. |
When a machine starts, it has to get its IP address some way or another. It could be a preset (hard-coded) adress, given by a DHCP server or set using some kind of peer-to-peer technology like rendezvous.
It is only when the machine has started and well established as a node in networked environment that it could start offering and/or looking for network services.
Here, Jini could start playing its role to help clients to connect to services and helping services to broadcast their existence and capabilities to potential clients.