From: MADMAN (dmadlan@xxxxxxxxx)
Date: Wed Feb 13 2002 - 19:05:14 GMT-3
Yes I agree with your findings but would elaborate a little. I might
be a bit jaded, fought several bugs with the MLPPP and the PPP overhead
and interleaving problems, I seem to recall interlaving would fragment
packets which caused problems, among them delay. I think there is a ppp
mutlilink do not fragment command which helped. I'm definately a CEF
bigot for it's simplicity, CPU friendliness, ease of configuration and
troubleshooting and fewer bugs.
Dave
Jason Gardiner wrote:
>
> I've been working some major issues with CEF/PPLB and MLPPP. Here's what
> I've been able to determine:
>
> MLPPP does interleaving that makes the bundle appear as one physical link.
> The packets are sent and arrive through one IP address.
>
> CEF/PPLB allows a per-packet load sharing through 2-6 ip addresses on each
> end.
>
> CEF/PDLS allows load sharing based on source/destination pairs. But any one
> flow is limited to the bandwidth of one link and the traffic distribution is
> not exactly equal.
>
> End operating systems have to handle packet reordering from multiple source
> addresses when using PPLB. Some clients, notably Microsoft, don't like this
> too much and the overall performance suffers.
>
> Of course, the router CPU overhead on MLPPP can be pretty high, depending on
> the platform.
>
> On Wednesday 13 February 2002 15:42, MADMAN wrote:
> > I know curiousity killed the cat but why? What would this provide
> > that can't be done with CEF besides not burning extra subnets.
> >
> > Dave
> >
> > "McCallum, Robert" wrote:
> > > Hi,
> > >
> > > Does anyone know when multilink frame will be hitting IOS for the 2600's
> > > and above? OR does anyone know where I can find out.
> > >
> > > Cheers
> > >
> > > Robert McCallum CCIE #8757
> > >
> > > "You can swim all day in the Sea of Knowledge and
> > > still come out completely dry. Most people do."
This archive was generated by hypermail 2.1.4 : Thu Jun 20 2002 - 13:46:22 GMT-3