Date: Thu, 08 Jan 2004 15:57:43 +0100 From: Andy Sporner <sporner@nentec.de> To: dingo <dingo@microbsd.net>, freebsd-cluster <freebsd-cluster@FreeBSD.ORG> Subject: Re: FreeBSD Cluster software/FREP Message-ID: <3FFD6FE7.9090109@nentec.de> References: <1072207597.3128.14.camel@vaio.isosweb.com>
next in thread | previous in thread | raw e-mail | index | archive | help
dingo wrote: >Two things: > >I saw the post on the freebsd clusters mailing lists but there seems to >be no code on the web site link in the message. > >http://www.sporner.com/freebsdclusters/frep > > There has been a change in the behaviour in the webforwarding. It used to be that all urls were passed to the new domain, which is in fact 'sporner.dyndns.org' (or sporner.dnsalias.org), but somehow it isn't done that way any longer. I will end up moving my domain to dyndns and this will be solved, but for the moment just use the following URL: http://sporner.dnsalias.org/freebsdclusters/frep >What it is: > >Well it's a start of a directory replication scheme. In the moment it >replicates changes from one filesystem tree to another on a remote >machine. NO THIS IS NOT NFS!! As files are changed, the changes >are reflected to another node. This creates a duplicate of the data >changes >that are occurring locally. > >and what is Optim/OS ?? i get a 404 Error on the link > This was an early (1996) idea that I had that never got past the drawing board about a minimalist kernel that only was a message switch and memory manager (in fact the message switch was to be build by doing page mapping rather than copying). The idea was to allow processors (or nodes) to be managed as a community resource that services could be dynamically instantiated (on demand) and since everything ran as a service they could be located anywhere, with the message layer being the conduit. I later realized that it made no sense to continue this work because of all the systems that exist it didn't make a lot of sense to create yet another one--except for ego reasons, which is not enough. So instead I got on the idea of process migration using an existing system, which is where things are now. In retrospect the first idea might not be a bad one when it comes to clusters of hundreds of machines, but I decided to go this way, which is where things will remain for now. Andy
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3FFD6FE7.9090109>