Re: How bandwidth requirement could be reduced when using thin clients?
I'm rather new to the list, so forgive me if I repeat things already
covered in the past.
Knut Yrvin wrote:
The municipal department of education in Oslo expect us to place the
application servers centrally for 175 schools. The idea is that all
the servers should be placed in one building, and they should rely on
heavily use of thin clients on a serverfram. At our first meeting they
told us that they had solved this with Windows and Citrix. In real
life it's like this:
Have you considered alternatives to LTSP? Owen LeBlanc at the University
of Manchester has recently deployed a few thousand Debian desktops there
using a combination of AFS (but any network file system could be used)
and a 4.5Mb local installation with only himself as the administrator.
Bandwidth costs as I understand it are not immense due to extensive
local caching of applications. Owen's current write up of his work is here:
Another possibility might be Nymph developed by Fensystems/Michael
Brown, it might best be described as a medium client solution where
machines over a certain BogoMIPS have some of the servers load shared
out to them and lower spec machines act as pure thin clients. We
deployed this at a school in Manchester last year and has been a
reasonable sucess, hampered mainly by politics.
The downsides to Nymph is that it is very Mandrake and Perl centered,
and works by exporting the root FS over NFS, therefore requiring a
number of tweaks to KDE and currently lacks a robust FS. Source RPMS can
be found here:
Still speaking as someone who has just deployed over 100 dual boot
systems at another manchester school using the traditional method, Thin
Clients suddenly look alot nicer :)
On two occasions I have been asked [by members of Parliament],
'Pray, Mr. Babbage, if you put into the machine wrong figures, will the
right answers come out?' I am not able rightly to apprehend the kind of
confusion of ideas that could provoke such a question.