[ Date Index ][
Thread Index ]
[ <= Previous by date / thread ] [ Next by date / thread => ]
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Okay, here's a boring and undirected post that I expect everyone will ignore, but maybe it'll give someone an idea at some point in the future: Heres the basic vision: * More powerful server * Less powerful workstations The clients run their own X servers, but run everything else on the server. This has the advantage that programs should start quickly, if the server has a fast disk or already has some of the files in RAM. But then, gradually, the processes migrate away from the server, probably towards the workstation, to better utilise the resources of the cluster as a whole, and possibly increase responsiveness slightly. On top of that, it would be pretty cool if the network was topoligically divided into segments and files were cached accross the workstations (on disk if they have them, or in RAM) and fetched from active neighbours if available, rather than the server. This idea would strike a balance between the "fat client + fileserver" and "fat server + thin client" schools of thinking and, IMHO, make better use of present-day computing power. I know the OpenMosix team have designed a kernel capable of doing the process migration, but does anybody know if there is a similar project to the LTSP that impliments this idea? Jon -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQE+m0RqeTVvFHAhe5cRAtUHAJ98Zp519Lce3qBb0DenLp901H4qvACfboYG kdwgaVcfbi6VTyJ79LblkK8= =PpGY -----END PGP SIGNATURE----- -- The Mailing List for the Devon & Cornwall LUG Mail majordomo@xxxxxxxxxxxx with "unsubscribe list" in the message body to unsubscribe.