[Jyri Virkki:]
| > and it has a link to the Squid Feature poll that has been setup.
| > http://vrthra.googlepages.com/squidpoll.html
| 
| I'm surprised some of these (like logging) must be chosen at
| compile-time, that's quite limiting. 

If the logging is disabled during compile time, It is removed completely
from the Request Path. (Introducing no overhead at all). While even if
the logging is enabled during compile time, The administrator can
disable it from the squid.conf file. (With a --possibly small -- overhead
for checking the squid.conf option)

| If the functionality is compiled in, is there any way to disable it
| through configuration (and if disabled, is the performance impact
| still the same)?

see above.

| Off the top of my head I think ident-lookups is not important (but I'm
| sure someone somewhere might care), but the rest of the listed
| features are often important.

These can still be disabled by the administrator if necessary even if
compiled in. So if pure performance is not paramount it may be better
for us to leave them on.

I see that CoolStack has disabled ident lookups,
http://wikis.sun.com/display/WebStack/SquidCache would Shanti be willing
to comment on this?

The other distribution that has it disabled is FreeBSD ports, which AFAIK
is geared towards good performance.

| I don't think it is possible to have a meaningful opinion without data
| on the performance impact of each option. Are we talking 25% worse or
| 0.5% worse or what?

Looking at the code where the options are #ifdef'ed it may be more of
the later figure, but I dont have any data on the relative performance
for these.

| Unfortunately the poll doesn't provide for a way
| to include supporting data for the choice(s). Are you aware if anyone
| has run controlled performance tests on these choices?

If we could find a place to host a cgi or servlet, it would be easy to
do this kind of stuff in the future (with places for supporting data on
choices). - as of now, it is limited by what the free polling sites offer
us.

                                    rahul
--
1. e4 _

Reply via email to