Mailing List Archive

At 10:46 AM 03-09-02 +0100, Mark Ivens wrote:


>On a related note, does anyone know when the NPE-G1 is going to arrive?

According to:
http://www.cisco.com/warp/public/cc/pd/ifaa/prossor/prodlit/npeg1_ds.htm
it is available already.

-Hank


>Cheers
>Mark
>_______________________________________________
>cisco-nsp mailing list real_name)s@puck.nether.net
>http://puck.nether.net/mailman/listinfo/cisco-nsp
>archive at http://puck.nether.net/pipermail/cisco-nsp/

Hi,

On Tue, Sep 03, 2002 at 01:10:44PM +0300, Hank Nussbacher wrote:
> >On a related note, does anyone know when the NPE-G1 is going to arrive?
>
> According to:
> http://www.cisco.com/warp/public/cc/pd/ifaa/prossor/prodlit/npeg1_ds.htm
> it is available already.

Well. Our sales contact still tells us "end of next month" (since about
4 months or so).

The URL quoted above lists 12.2(8)B or 12.2S as minimum IOS requirement,
neither of which is available on CCO...

gert
--
USENET is *not* the non-clickable part of WWW!
//www.muc.de/~gert/
Gert Doering - Munich, Germany gert@greenie.muc.de
fax: +49-89-35655025 gert.doering@physik.tu-muenchen.de

Thus spake Gert Doering (gert@greenie.muc.de):

> The URL quoted above lists 12.2(8)B or 12.2S as minimum IOS requirement,
> neither of which is available on CCO...

Although 12.2(4)BW is, a special release with support for the NPE-G1.

http://www.cisco.com/warp/public/cc/general/bulletin/software/ios122/1877_pp.htm

> A previous post on this list cited 1500 L2TP sessions as a real world
> limit on an NPE-300 LNS.
>
> Now that L2TP is PXF accelerated in 12.2(4)B, what real world
> performance improvements (if any) could be expected between an NPE-400
> and an NSE-1? For example, Cisco quotes a 65% improvement for PPPoA
> (rather than L2TP) but this is in the same FAQ claiming an NPE-400 can
> support 8000 sessions :-)

Note that this depends a *lot* on how many session connects/disconnects
you have per second. We found this out the hard way when trying to use
NPE-300 for L2TP termination of dialup customers (typically 2 x ISDN
BRI). Our boxes were getting killed by the signalling, not the actual
L2TP packet shuffling.

If you're doing PPPoA for ADSL customers, your signalling rates will be
much lower, of course.

Steinar Haug, Nethelp consulting, sthaug@nethelp.no

Also, you must be very careful with features on PXF - if you get it
wrong, you will shut off the PXF processing.
I know about problems with MSS Adjust on several platforms/versions.

Arie

-----Original Message-----
From: sthaug@nethelp.no [mailto:sthaug@nethelp.no]
Sent: Wednesday, September 04, 2002 10:43 PM
To: mivens@clara.net
Cc: cisco-nsp@puck.nether.net
Subject: Re: [nsp] Real world L2TP PXF performance increases


> A previous post on this list cited 1500 L2TP sessions as a real world
> limit on an NPE-300 LNS.
>
> Now that L2TP is PXF accelerated in 12.2(4)B, what real world
> performance improvements (if any) could be expected between an NPE-400

> and an NSE-1? For example, Cisco quotes a 65% improvement for PPPoA
> (rather than L2TP) but this is in the same FAQ claiming an NPE-400 can

> support 8000 sessions :-)

Note that this depends a *lot* on how many session connects/disconnects
you have per second. We found this out the hard way when trying to use
NPE-300 for L2TP termination of dialup customers (typically 2 x ISDN
BRI). Our boxes were getting killed by the signalling, not the actual
L2TP packet shuffling.

If you're doing PPPoA for ADSL customers, your signalling rates will be
much lower, of course.

Steinar Haug, Nethelp consulting, sthaug@nethelp.no
_______________________________________________
cisco-nsp mailing list real_name)s@puck.nether.net
http://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Thus spake Arie Vayner (ariev@netvision.net.il):

> Also, you must be very careful with features on PXF - if you get it
> wrong, you will shut off the PXF processing.
> I know about problems with MSS Adjust on several platforms/versions.

Would you mind saying which platforms/versions?

For 12.2(4)B4 the only relevant defect I can see on CCO is CSCdw93170
"PXF still punts packets after removal of ip tcp adjust-mss" which I'm
not too bothered by since I'll be leaving tcp adjust-mss on.

On Tue, 10 Sep 2002, Mark Ivens wrote:

I know for a fact that this version has problems on 7400. We got some
special version for that, but did not test it yet.

Arie

> Thus spake Arie Vayner (ariev@netvision.net.il):
>
> > Also, you must be very careful with features on PXF - if you get it
> > wrong, you will shut off the PXF processing.
> > I know about problems with MSS Adjust on several platforms/versions.
>
> Would you mind saying which platforms/versions?
>
> For 12.2(4)B4 the only relevant defect I can see on CCO is CSCdw93170
> "PXF still punts packets after removal of ip tcp adjust-mss" which I'm
> not too bothered by since I'll be leaving tcp adjust-mss on.
> _______________________________________________
> cisco-nsp mailing list real_name)s@puck.nether.net
> http://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>

Thus spake ml@vayner.net (ml@vayner.net):

> I know for a fact that this version has problems on 7400. We got some
> special version for that, but did not test it yet.

Do you have a bug ID for it?

Mark Ivens <mivens@clara.net> writes:

> Thus spake Arie Vayner (ariev@netvision.net.il):
>
> > Also, you must be very careful with features on PXF - if you get it
> > wrong, you will shut off the PXF processing.
> > I know about problems with MSS Adjust on several platforms/versions.
>
> Would you mind saying which platforms/versions?
>
> For 12.2(4)B4 the only relevant defect I can see on CCO is CSCdw93170
> "PXF still punts packets after removal of ip tcp adjust-mss" which I'm
> not too bothered by since I'll be leaving tcp adjust-mss on.

Enabling MPLS on the GE disables PXF switching even for those packets
which are not MPLS-switched.

An issue with 12.2(4)B4 is that it didn't free vaccess interfaces,
which why we had to revert to 12.2(2)DD4.

Robert