What Happened to Jini?

JJ asks: "What has happened to Jini, Sun's networked mobile code technology that debuted with Mmuch fanfare nearly two years ago? The answer seems to be not much. What's the reason for the lack of adoption of Jini? Is it because nobody can understand the trademark and licensing requirements? Or is it because few people have realized Jini's true potential?"

JINI is way before its time...

[Trak]

I think the problem is more a matter of "who needs it?" We don't have an immediate need for our refridgerator to talk to our printer to talk to our garage door opener. Sure it's cool to plug a network cable into the back up a printer and have everybody suddenly be able to print to it... but do we need that level of convenience? Nope. It's simple to set up network printers on your typical windows or mac network. Microsoft's "plug and play" is plenty adequate for the current consumer needs.

Re:Not dead, but dying

[Anonymous Coed]

You dork, jwz didn't work for Sun, he worked for Netscape! jwz has nothing to do with Java / Jini!

Re:Not dead, but dying

[Anonymous Coed]

Umm, what exactly about that document is proof? He talks about working in Netscape there.

Re:It hasn't gotten adopted

[Bazzargh]

re the Palm, wasnt there a JavaOne demo of the KVM (the small footprint jvm for the palm etc) which involved it controlling Mindstorm robots? While the KVM (and its source) is downloadable, theres no sign of any Jini implementation for the Palm. The article about that demo is here:
http://developer.java.sun.com/developer/technica lArticles/ConsumerProducts/JavaTanks/PalmP roxy.html

...and while it actually does have all the classes for the demo, if you look closely you'll find the PalmProxy refers out to a half-dozen net.jini packages (not supplied), and rmi packages (ditto). The KVM is pretty limited (no JNI, serialization, lacks coverage of the Palm API) and I'm not convinced the standard Jini/RMI packages would work here.

I did find a palm jini/rmi implementation though, after a little searching:
http://postpc.cs.berkeley.edu/rmilite/

Re:JINI is way before its time...

[lw54]

Microsoft's "plug and play" is plenty adequate for the current consumer needs.

But this is exactly the reason why Jini was developed. Some people don't like Microsoft "making things easy" for everyone. :-)

Jini is amazing.

[davesag]

My guess is that the people who are currently deploying stuff using Jini are so damn excitied by the faster development times that it offers that they don't want others cutting into their lead. Jini is not just about your printer talking to your toaster, it's a fantastic set of patterns for building almost anything. We are using Jini and JavaSpaces particularly and the rewards have been amazing. When the first Jini storage machines and printers come out, connecting our backend code to them will automagic. Until that time, we still get the advantages of our backend system being faster to replicate, deploy and scale than any other system I have ever worked with.

cheers

dave

Re:JINI is way before its time...

[AbbyNormal]

Plus (as the webpage states), Microsoft doesn't want me to have it my way! RIiiiiiiiiiiiight.

All Hail Bill. The pointy haired one has spoken.

M$ bashing can only go so far (but its fun).

jini over bluetooth

[mckwant]

Heard today that BlueTooth chips will be $5 in volume by next year sometime. I suspect that a big problem w/Jini was that you needed to be on a network to use it, and most refrigerators, AFAIK, don't have rj-45 jacks.

Wireless AND ubiquitous access to the net, THEN they'll have something functional. What they'll do with it after that, I frankly have no idea. A "contents of my fridge" web page would be nice, but why? Now, remote control of my stereo from anywhere in the house....

wouldn't a Beowolf cluster of these be cool?

Where's the market?

[anacron]

In order for JINI to become a success, every device needs to implement the same spec -- a spec they can't help shape. So what if you're the first JINI enabled device? ... You need a second device and maybe a third for your product to become useful. On top of that, you also need the infrastructure to provide the connectivity. So what can JINI do that a Microsoft server can't? Printer sharing? Already done. Information exchange? Maybe .. but why would my VCR care what my refrigerator's temprature is?

I think the value of JINI is not that it enables devices to be connected together ... but that it would provide a way to monitor all those connected devices from a central (or many) locations. The 'SmartHouse of the future' doesn't rely on the fact that everything's networked ... it relies on the fact that everything can be managed from as many locations as you can think of.

I see value not in being able to connect devices together, but in putting web servers inside everyday products. Once your house is networked, you plug the device into the cat-5 outlet and hit an IP address -- the device itself provides the web server and interface.

Licensing the operative factor

[AmateurPhlebotomist]

The insane licensing requirements are where, IMHO, the bottleneck is occurring. Since one cannot sell a jini service without negotiating a license agreement with sun, there will be no rich third-party market of services. Without that, no 'critical mass' of services/applications to give incentive for widespread adoption.

When I read the licensing agreement, it appeared that one would have to negotiate an agreement with sun even to distribute a jini service for free. When I saw that, I just went ahead, put the jini and javaspaces books away, and moved onto the next topic on the crowded tech topics queue.

The answers so far say it all....

[X]

The major problem with Jini is how it has been marketed. Most of the answers to date talk about the problems using Jini for things like refridgerators and such. There's nothing about Jini that makes it an "appliance" technology. Despite Sun's marketing to the contrary, Jini is basically a mechanism for supporting dynamic, ad-hoc network applications.

The sad reality is Jini is actually an excellent technology for enterprise solutions (particularly web solutions). Imagine the benefits of having enterprise services which can be discovered and used automatically. This means that when a system gets loaded, you just deploy a new Jini-tized component, and the without any configuration changes, the software starts spreading the load over to that component.

Unfortunately, Sun didn't tout this aspect of Jini. Indeed, 90% of the people who know what Jini is think it's a technology for connecting network devices. That's a very limited and unproven market. I suspect Sun didn't want to let J2EE's image become confused by allowing Jini to move in to that space.

That in and of itself wouldn't have been a fatal blow, but on top of this was the licensing problem. Jini's license made it hard to deploy Jini solutions without paying Sun some money. Particularly given that the technology was far from perfect and still needs a several evolutions before it'd work right, 3rd parties weren't too interested in overcoming these problems just to see their solutions become Sun property which could be licensed to their competitors.

Finally, Sun's ability to corral vendors together was starting to wane just as Jini was introduced. Sun had been the standard bearer for the anti-Microsoft camp with their Java technology, but by the time Jini was being promoted Sun was already starting to get a negative image from how they were handling Java. It's hard for a single company to maintain the position that Sun had for the preceeding 3 or 4 years.

So, in summary, Jini was the right technology introduced by the wrong vendor, at the wrong time, in the wrong way, and targeted at the wrong market. :-(

It hasn't gotten adopted

[maggard]

Jini has the classic chicken-and-egg problem: until somebody uses it nobody wants it.

How to break that cycle? Well, there are a few ways:

1. Solve a well defined problem. Jini doesn't do this. It can do a lot of things but it's hard to express in the proverbial one sentence.
2. Become ubiquitious through bundling with the OS. Unfortunately this would mean inclusion with Windows (not gonna happen.) Furthermore Jini indirectly competes with MS's own Active Directory (& Novell's NDS) so it's not going to win even tacit support.
3. Be trivial and safe to install. Or at the very least be freely downloadable and easily installable. With the cumbersome licensing and convoluted installation process required this isn't happening.
4. Come included in the hardware. If HP, Xerox, IBM, LexMark, Agfa, Palm, Nokia, Motorola, and other makers of networked or networkable devices were to start including Jini then it'd stand a chance. However with Sun's licensing and lack of momentum this seems unlikely. At this point Sun would have to supply these vendors with quality code and engineering support for customizing - something Sun doesn't seem to appreciate or be able to do.

Sure - Jini could be a great feature. It would allow you to walk onto a campus (educationial, corporate, whatever) and get access to all of the devices publically availiable. It would allow IS folks to just pop devices down wherever they're needed and do a minimum of configuration. Moving to a new building? With Jini-enabled printers, desktops, laptops, scanners, door-locks, phones, etc. it could be done overnight.

With Bluetooth and wireless-LANS finally becoming a reality all of Jini's features become that much more attractive and pressing. Even the US DOD drools over this kind of ability (set up a battlefield HQ in a few minutes) though their process is so glacial and baroque that nothing is likely to result from their support that would apply to the rest of us.

The really suprising thing is that Sun doesn't seem to be evangelizing Jini at all. It's possible they can't figure out how to apply this kewl technology gifted to them (they paid for it but it seems to have pretty much come out of the blue to most folks at Sun.) I've not even noticed any technology demos of it. Is anyone aware of anything outside of a lab?