Grid Computing with Plan 9 Ants(9gridchan.org)
9gridchan.org
Grid Computing with Plan 9 Ants
http://www.9gridchan.org/
15 comments
If anyone is curious about what this is all about, feel free to ask, this is my project which has been ongoing in one form or another since 2008.
Hi mycroftiv! I remember when you started all this in #plan9 and #cat-v and I am very glad of seeing you are still running the grid.
At some point, I could not follow your pace and stopped following. I was wondering, what interesting use you found that was totally unexpected? There were so many open possibilities last time I checked that I am sure you found something nice in there.
At some point, I could not follow your pace and stopped following. I was wondering, what interesting use you found that was totally unexpected? There were so many open possibilities last time I checked that I am sure you found something nice in there.
Thank you, I remember your participation back in that era. The main useful thing that emerged somewhat recently is the system called 'spawngrid' which leverages fossil rootscores, venti replication, and independent namespaces provided by ANTS to spawn independent plan 9 environments on demand and make them available on multiple nodes. It was a bit too ambitious for the real world so I'm going to stop wasting the money on running the unused servers but it did work pretty well in the technical sense. The lighter and simpler public grid will continue because it has an active user community.
I also developed a lot of interest in more theoretical computing topics which I may or may not be able to bring toward Plan 9 namespaces at some point, but I don't have the time to summarize that here, the blog on my site covers some of the ideas.
I also developed a lot of interest in more theoretical computing topics which I may or may not be able to bring toward Plan 9 namespaces at some point, but I don't have the time to summarize that here, the blog on my site covers some of the ideas.
OK, what is this about?
It's about two things mainly: updating the design of the distributed architecture of Plan 9 to be less centralized, and using Plan 9 for ad-hoc resource sharing and communication for hobbyists. These things overlap because I provide grid resources for people to use which run on ANTS-powered servers/architecture.
> send auto-opening web links
Is this wise?
Is this wise?
No, its not wise from a security perspective, the grid operates in a very fragile zone of 'nobody except a small number of plan 9 users knows or cares about this and there is no incentive for malicious behavior' - we have been very lucky in that it seems that using Plan 9 and taking the time to learn how to connect to the grid seems to filter out people who just want to be destructive or malicious. At some point it may prove to be a problem but we have gotten along ok for a few years. Users also have the choice of what services they use and some users dont use the plumber for auto-sharing in this way.
what are some unique things people on the grid do?
It's mostly hobbyist Plan 9 software development. There is some shared disk storage and people upload code they are working on for other people to test and help with, so it is an active workspace for Plan 9 development. There is also a lot of general discussion of computer related topics in the chat, sometimes with a rather abstract mathematical focus on limits of computation and dependent type theories and the logic of infinity, with only a vague connection to existing Plan 9 software - far out brainstorming.
Do some people have specific ideas or visions of what this could turn out to be? I've tried to play this out in my head but wasn't able to arrive at any new paradigms but then again I've only scratched the surface of plan 9 grid
I've done some experimental setups using hypercubic routing and algorithms inspired by 80s supercomputer designs from the nCUBE corporation, and the 'spawngrid' is a lighter-weight plan 9 take on containerized service architecture, so I'd say the answer is yes.
FYI spawngrid is getting mothballed as there are no current users so there's simply no reason to keep it running.
The ants distribution is also "discontinued" as it is a downstream fork of 9front. mycroftiv is a one man show and it was too much work to keep patching his kernel to stay in sync with upstream. Though you can grab the patches and get them working on 9front if your willing. (send patches and all that)
Though the gridchat is a very active, friendly place to talk. Lots of knowledgeable people on there. Grab the gridstart and chat scripts and run it on 9front or labs plan 9. You can also get the chat script running on plan9port.
The ants distribution is also "discontinued" as it is a downstream fork of 9front. mycroftiv is a one man show and it was too much work to keep patching his kernel to stay in sync with upstream. Though you can grab the patches and get them working on 9front if your willing. (send patches and all that)
Though the gridchat is a very active, friendly place to talk. Lots of knowledgeable people on there. Grab the gridstart and chat scripts and run it on 9front or labs plan 9. You can also get the chat script running on plan9port.
Offtopic, but I've always wanted to abuse ants' (the insect's) ability to find the quickest path in a grid, to make a tiny calculator or memory array, redstone style. Thought for a moment that you had done that, and got myself excited
Are there any plans/possibilities to "upstream" this to 9front (or wherever is appropriate)?
From the discussions I've had over the years with the primary 9front maintainer, the balance of adding new features and complexity and the fact that ANTS is most effective in combination with the Fossil file server which base 9front removed from its distribution means that it goes in a different direction than 9front's goals. Some software which is part of ANTS such as 'hubfs' (9p muxfile/message queue) may be integrated into 9front via a different author's rewrite as 'mq' and some other ideas from ANTS may eventually result in 9front code and changes, but the ANTS patchset for 9front was around for several years and didn't attract enough users to create any groundswell of demand for integration into the base.