Hello all!
Common wisdom has it that NFS, in addition to other shortcomings, does not
scale very well, and that for large sites AFS (or DFS, given the right OS...)
is preferrable. My question is how to gauge the scale at which NFS (and NIS,
and auotmounter, and related stuff) starts to choke (without making a full
scale test, then). The scale is likely determined by things as bandwidth of
the network, of the busses of the servers, memory configurations of clients
and servers as well as sw configuration issues. The discussions I have seen on
this are mostly a few years old, and since then at least hardware has evolved,
and possibly the NFS code too. So I'd like to solicit some comments from
people with real world experience of large NFS installations.
At our institute one could expect about ten some servers exporting together a
few hundred mount points to hundred some clients, each client being fairly
small.
Any takers?
Regards,
Simon.
Simon Tardell, voice +46 8 162688 fax +46 8 347817
Fysikum, Stockholms universitet simon_at_physto.se, simon_at_tardell.se
Received on Tue Jan 13 1998 - 11:30:09 NZDT