NFS scalability

[Michael Neumann]

Chris Turner schrieb:
steve wrote:
On Mon, Sep 10, 2007 at 07:49:51AM -0700, Matthew Dillon wrote:
:Just out of interest (without the time/boxes to try it out myself), do 
:you guys think this is a "good" solution? I mean, does NFS handle a lot 
:concurrent accesses well? I never used NFS...
:
haven't played around with it,
but you can do interesting things with amd(8) maps to distribute the
load, handle failure, etc
I think this would be best suited for 'static' into (e.g. non-time
critical html, php scripts with only logic, etc) - as Matt points
out there are cache problems on NFS (have seen this occasionally first
hand when serving up NFS builds on a lan)
Thank you all for the informative answers.

In the setting I am thinking about, serving big static files would be 
the top priority. Those files would hardly ever be updated.

But on the other hand tunneling all traffic from the front-end 
web-servers via NFS to the storage server back and forth seems overkill 
to me. Ideally each storage server would transfer the content it owns 
directly to the user instead of tunneling it through a proxy. But I 
think this must be done at the application level :)

Regards,

  Michael