On 1/23/06, Charles Sprickman <[EMAIL PROTECTED]> wrote:
> Just as another datapoint, I had 640kb/s in for a 768kb/s adsl upload.
> I've backed it down to 600kb/s, which is probably overkill.  SSH still
> gets laggy when a bulk upload is going on.  VoIP MOS score goes from 4.3
> to 3.0 with a bulk upload as well.

Don't forget that the 768K that you're provider is telling you
includes the transport overhead.  PPPOE and ATM frames do eat up a
non-zero amount of bandwidth.  The wizard was (and now does) supposed
to remove 20% from whatever value you give it which puts you at 614K. 
20% loss seems a little extreme, but it's not horrible when your SSH
sessions barely slowdown during a download/upload.  Part of the 20% is
also so that it can buffer traffic bursts before forwarding on.

Best results will come from pfSense at both sides of a link
controlling traffic flow, but it does a pretty good job even with just
one leg.

> Also, I'll ask my bpf question again:  Is there a simple way to get bpf
> enabled on a pfsense box so I can do tcpdump analysis directly on the
> firewall?  If I had a pfsense kernel config file, I assume I could just
> build a kernel elsewhere and move it over, right?

I didn't see the first question, but tcpdump works just fine on
pfSense (we use it for pulling the logs off pflog0).  And I'm
frequently making use of it for troubleshooting.

--Bill

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


Reply via email to