> If I'm going to implement for 500 Users , how many
> ntlmauthenticator processes I should specify in
> squid.conf
Hard to say - each individual setup is different.
The general rule of thumb is one NTLM auth helper for
each user.
At the same time, our network has over 20 concurrent users
and we use only 10 helpers (and the first three handle
over 75% of the requests).
I would suggest using Cache Manager to keep an eye on
your NTLM auth helpers. It will show the number of requests
handled by each helper, and the number of seconds since each
helper served a request. There's also statistics for the
current queue length and the average service time.
In addition to increasing the number of helpers, you can
also increase the max_challenge_reuses and
max_challenge_lifetime parameters.
The problem might not lie with Squid - if the DC is busy,
or the link is congested, that can cause delays that will
make requests queue up.
Adam
Received on Tue Aug 19 2003 - 06:25:38 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:18:59 MST