Howardism: Whence Jini?

Update

Before the flames come, let me say that Jini isn't dead … it has an active community and whatnot. What I meant is now that the marketing hype from Sun is gone, I don't hear developer-friends mentioning it much and this got me thinking as such.

Also I should state that I know that I'm not technically accurate in my posting … Most of my readers (both of them) are … well, they have the attention span of a gerbil on hashish, and no one seems to finish reading my posts if they have to reach for the scrollbar.

Granted, I never want to succumb to the Siren's soundbite that plagues many a weblog. So, I compromise … now, on to my 5th Symphony.

Update

As of 2005, Jini has a new licensing model. Sun originally released Jini under the almost open-source, Sun Community Source License (SCSL), but now opened the licensing model and began releasing their specifications and implementations under the Apache license.

I wonder if this will help it gain in popularity.

Whence Jini?

The problem with marketing hype in the high-tech industry is that little can live up to such high expectations. A number of good ideas seem to be forgotten purely because of a lack of understanding of what the idea may have been actually good for. One of my personal favorites is Jini.

When it was first hyped up it was hailed to revolutionize the personal electronics industry as well as just about everything else. But it's strength was in its practical approach to distributed computing. And now that Grid Computing is now getting popular, I can't help but wonder what happened to Jini. Of course I can't blame its lack of adoption solely because of its publicity as there were other factors.

We just don't think in distributed ways.

I mean, we mostly think serially and most of our early computer languages were made to execute instructions sequentially. If you solve problems with a divide-and-conquer approach, you come up with computer languages that are more mathematical and functional (like Lisp).

Ever since the Roman legions, we've been dealing with hierarchies and this shows up in the Unix file system (popularized by DOS and the Macintosh folder system). Systems like XML and HTML fall into this category.

But we don't often solve problems in a distributed fashion… well, not in a heterogenous distributed fashion. Now my off-handed comment is getting the place sticky …

We often use homogenous distributed systems, commonly known as clusters, where every "node" acts like every other node. The most popular web sites are hosted in such a way. But systems where each node can be unique are harder to find. Yes, an n-tier application framework is often distributed on unique nodes, but the problem it solves is usually serial.

The most common example of a distributed system is the web. Each node can be truly unique in the data it submits to every other node in the Internet system. So perhaps "intranet solutions" like corporate information portals might be solved well with such distributed systems. But Jini might be overkill in such a situation.

Jini's advantage is when a system requires some sort of guaranteed action from multiple nodes. With it, you can have a transaction that either works on all the nodes or doesn't work at all. But typical "portal" solutions are set up so that if a node offers data, it gets retrieved by the portal. The data usually doesn't require more cooperation that this.

Another feature of Jini (that can be found in many other systems) is its network registry, where when a node with a service can submit details about itself to the system at large. Portals could use this as a way to discover new "feeds" of information.

Maybe …

I don't have any solution or proposal here … just musing. I always enjoyed reading and toying around with Jini (never built anything real with it), but I liked it and hope that it might make a come-back of sorts… but without all the hype. I generally don't trust things with lots of hype.

Tell others about this article:

Click here to submit this page to Stumble It


	Java Not Fun? Freemarker vs. Velocity Loosely Coupled Components, Part 1 Computer Inspiration What's up with Java? Thoughtful Consideration better than any Pattern Interview Questions Why Not Use the Database? Perspectives on Struts Read other entries in Java...	Update Before the flames come, let me say that Jini isn't dead … it has an active community and whatnot. What I meant is now that the marketing hype from Sun is gone, I don't hear developer-friends mentioning it much and this got me thinking as such. Also I should state that I know that I'm not technically accurate in my posting … Most of my readers (both of them) are … well, they have the attention span of a gerbil on hashish, and no one seems to finish reading my posts if they have to reach for the scrollbar. Granted, I never want to succumb to the Siren's soundbite that plagues many a weblog. So, I compromise … now, on to my 5th Symphony. Update As of 2005, Jini has a new licensing model. Sun originally released Jini under the almost open-source, Sun Community Source License (SCSL), but now opened the licensing model and began releasing their specifications and implementations under the Apache license. I wonder if this will help it gain in popularity. Whence Jini? The problem with marketing hype in the high-tech industry is that little can live up to such high expectations. A number of good ideas seem to be forgotten purely because of a lack of understanding of what the idea may have been actually good for. One of my personal favorites is Jini. When it was first hyped up it was hailed to revolutionize the personal electronics industry as well as just about everything else. But it's strength was in its practical approach to distributed computing. And now that Grid Computing is now getting popular, I can't help but wonder what happened to Jini. Of course I can't blame its lack of adoption solely because of its publicity as there were other factors. We just don't think in distributed ways. I mean, we mostly think serially and most of our early computer languages were made to execute instructions sequentially. If you solve problems with a divide-and-conquer approach, you come up with computer languages that are more mathematical and functional (like Lisp). Ever since the Roman legions, we've been dealing with hierarchies and this shows up in the Unix file system (popularized by DOS and the Macintosh folder system). Systems like XML and HTML fall into this category. But we don't often solve problems in a distributed fashion… well, not in a heterogenous distributed fashion. Now my off-handed comment is getting the place sticky … We often use homogenous distributed systems, commonly known as clusters, where every "node" acts like every other node. The most popular web sites are hosted in such a way. But systems where each node can be unique are harder to find. Yes, an n-tier application framework is often distributed on unique nodes, but the problem it solves is usually serial. The most common example of a distributed system is the web. Each node can be truly unique in the data it submits to every other node in the Internet system. So perhaps "intranet solutions" like corporate information portals might be solved well with such distributed systems. But Jini might be overkill in such a situation. Jini's advantage is when a system requires some sort of guaranteed action from multiple nodes. With it, you can have a transaction that either works on all the nodes or doesn't work at all. But typical "portal" solutions are set up so that if a node offers data, it gets retrieved by the portal. The data usually doesn't require more cooperation that this. Another feature of Jini (that can be found in many other systems) is its network registry, where when a node with a service can submit details about itself to the system at large. Portals could use this as a way to discover new "feeds" of information. Maybe … I don't have any solution or proposal here … just musing. I always enjoyed reading and toying around with Jini (never built anything real with it), but I liked it and hope that it might make a come-back of sorts… but without all the hype. I generally don't trust things with lots of hype. Tell others about this article:
	2014, Howard Abrams Some rights reserved	Originally written on 22 March 2003. Updated on 8 February 2014.