Why has JXTA been abandoned? Any alternatives out there?

Question

P2p/Grid Computing seem like a promising concepts. JXTA looks like the only all in one framework for it. Is there a reason this field is so sparsely pursued?

Fumikofumitory

Answer 1

I have lead the release of JXTA 2.6 and 2.7 - JXTA is not completely abandoned. Some people have posted patches on the 2.6 branch and it could easily be merged with the 2.7 branch.

There are many reasons why people did not carry on participating to JXTA:

1. Oracle did not follow-up on their duties regarding project governance, which left the project in a limbo state.

2. Oracle did not follow-up on a request to move the project to Apache.

3. The code base was old. We cleaned it and implemented unit tests. But in order to move the project to the next level, it would have required a lot of rewriting. Not enough volunteers.

But more fundamentally, the reason few P2P frameworks took off is because P2P is fundamentally complex when you get into details. Most people don't get it until they start putting their hands in the dirt. It is not possible to implement P2P 'in a simple way'.

So nothing to do with all-Java clients, licensing fees or others.

Update (August 2013): You thought JXTA/JXSE was dead? Well someone worked further on it and developed a DZone tutorial (unfortunately, SO does not allow links to Dzone, so Google: JXSE and Equinox Tutorial).

Update (November 2013): A group of people is working on new releases of JXTA. For more information, register on the mailing lists.

Wellintentioned

Wow, thanks for taking the time to answer. I'm aware of your book on Scribd and have looked into getting my hands dirty with JXTA, but the dead links on the web and lack of documentation makes me think that I'll be wasting my time. I'm anxious to see P2P computing come together as a computing platform combining CPU & file sharing. It seems like the next logical step with CPU clock rates quickly approaching their theoretical limit. – Fumikofumitory

There are possible solutions to implement P2P and that's why I started the Chaupal project (on Google Projects), but I did not have the cycles to work on it. – Era

Answer 2

Interestingly what was missing with all the P2P initiatives of the past was a motivation for a peer to stay active. Question always was why would a peer keep running a CPU draining and XML based verbosed protocols. Trust was another factor - how can I trust a peer. As a key member of the team, we introduced security. But security doesn't address trust.

To make it even worse, JXTA introduced the concept of super nodes - defeating the very concept of peer to peer.

However, not everything was that bad. JXTA provided a lots of new concepts. One being Edge Computing with JXME and JXTA sitting together - you can call it to be current day Fog computing where heavy lifting was on the JXTA node and some intelligence on the constrained JXME nodes.

Fast forward, Blockchain addressed gaps with addressing most if not all the questions that any P2P platform could not answer: trust, incentivizing peers, tamper proof and much more.

P2P is still alive :)

Tournai

Answer 3

I think it's for the same reasons that RMI, CORBA, and Jini aren't much in favor: complex and closed.

Simple and open win most of the time.

It might have had something to do with all-Java clients or licensing fees or something else.

It could be competition. MPI is a widely accepted messaging standard for computing. Hadoop is getting a lot of traction.

UPDATE: The answer that was accepted discusses why people may or may not choose to participate in JXTA. I think my answer has more to do with user adoption, which is different. Mine go back to the origins of JXTA, not the details of releases 2.6 and 2.7.

Kistner

Answer to the update: There are surely areas where JXTA could be simplified, but, there are also areas that JXTA (and most -if not all- P2P frameworks) do not cover properly. These would require a minimal set of functionalities which are not simple to implement and which cannot be made small or tiny. You can't implement a P2P framework on MPI only. – Era

MPI is heavily used in scientific computing. Maybe P2P isn't nearly the magic bullet it was made out to be. Is JXTA unjustly ignored, or has the marketplace spoken? Sony Beta was supposed to be superior to VHS, but it's gone. – Kistner

I wonder about this myself. For file sharing P2P is unmatched. Not sure the same can be said of CPU sharing, although, BOINC, SETI@home & BitCoin seem to work extremely well in their respective niches. – Fumikofumitory

Answer 4

If you work with Linux, try this: http://www.p2pns.org/ "P2PNS (Peer-to-Peer Name Service) is a distributed name service using a peer-to-peer network. The current focus of P2PNS is to provide a secure and efficient SIP name resolution for decentralized VoIP ( P2PSIP)." In most cases Name Resolution is enough to build up a P2P-App on top of it.

Heterogynous