Skip site navigation (1)Skip section navigation (2)
Date:      09 Nov 1998 15:01:19 -0500
From:      Chris Shenton <cshenton@uucom.com>
To:        dave <dave@comsite.net>, archie@whistle.com
Cc:        questions@FreeBSD.ORG
Subject:   Re: mpd + EQL, Livingston PM2 MLLB, alternatives? mpd load high?
Message-ID:  <86u308vdo0.fsf@samizdat.uucom.com>
In-Reply-To: dave's message of Mon, 9 Nov 1998 10:41:26 -0600 (CST)
References:  <Pine.BSF.4.02A.9811091026020.15534-100000@bsdserve1.comsite.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Archie <archie@whistle.com> replied:

> Brian's user-land ppp now supports MPP as you know, and he is
> more active in maintaining it, if you plan to have a lot of
> questions & problems :-)

I never *plan* on having problems! :-)  Thanks for the quick response
-- I just wanted to get a feel for how actively mpd was being
maintained, whether I should put my effort into user-land ppp instead,
etc.  I'll continue working on mpd for at least a while since I've
already gotten over part of the learning curve... I'm a pppd guy
normally so user-land ppp is yet another config to learn.


> I think gated should pick up mpd's static route changes.. though
> you may need to configure gated to better "respect" static routes..?

I have no problem on the homenet side with the default route, it's at
the ISP end where I'm losing. I believe you're right that gated must
not be picking up mpd's interface and therefore not getting the static
route. I'll have to figure out what to look for in the logs then learn
how to persuade it to do the right thing. Thanks.



dave <dave@comsite.net> replied:

> The patch was indeed for the PM2's multi-line load balancing, although it
> doesn't get 100% of the performance it could get.  The hack i submitted to
> mpd was to make it so that mpd simply alternates sending packets between
> the two modems, however, I definitely did see a good performance
> improvement over a single modem with the 2 couriers I was using.
> 
> I don't know if the hack will work with PM3s...I don't see any
> reason why it wouldn't, but I have never tested it.

OK, great. The PM3s will do real multilink PPP; the PM2 analog ports
didn't have enough memory to do real MLPPP, to maintain state of
fragments and all, according to MegaZone and the Livingston list. This
should at least allow me to get more throughput, even if not optimal,
on the POTS PM2.


> You will definitely get better performance with mpd on either end and
> going with multilink ppp, but if that is not an option, MLLB should get
> decent performance.

Well then I'll give this a shot, since I already have the box in
place. Should probably upgrade to 2.2.7-stable on the dial-in box and
mpd-1.0b5 on both, first...


Thanks a bunch.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?86u308vdo0.fsf>