Todd Vierling writes:

> : As far as I know, the P15 does synchronous MPP to the remote end, and
> : converts this into standard (single line) async PPP before sending it to
> : the host. So it shouldn't matter if the host supports MPP or not. (Don't
> : take my word on this, though - I haven't actually used the thing
> : myself).
> 
> Probably true, and I believe the Motorola BitSURFR has this ability
> too.

Yep, I checked with one of the developers to verify this is true for the
P15. As far as the BitSURFR, I think only the BitSURFR Pro has MPP.

>  It does some trickery with the PPP to make it work across multiple
> channels.

Although it isn't exactly accurate, I like to think of it as having a
router with 2 interfaces, one speaking sync PPP on ISDN and the other
doing async PPP on RS232. (The inaccuracy is that the packets aren't
completely decapsulated to straight IP inside the router, for reasons
that become obvious in a minute).

> : As far as STAC compression, I could be wrong about this as well, but the
> : information I have indicates the P15 doesn't support STAC compression
> : under any circumstances, regardless of host OS.
> 
> This only has to do with the PPP implementation of the host, actually.  The
> STAC compression is another layer atop PPP, after the channels have been
> split and reassembled by the P15.  The raw data sent to the serial port by
> the P15 is decompressed in the computer.

I checked with the devloper on this too, and you are correct (this,
along with plain old efficiency, is why the packets aren't completely
decapsulated when converting from sync to async). It also transparently
(except for escape sequence handling) passes through all the CCP
(Compression Control Protocol) packets in both directions so the remote
end and the local host can properly negotiate.

So, while the P15 supports MPP with *any* local host regardless of
whether the host itself supports MPP, the (lack of) support for STAC
compression is dependent on the local host PPP's support for STAC (or
whatever other compression algorithm.)

> : I agree with this whole paragraph (although I don't know anything about
> : dumping the P25). Async<->Sync ISDN terminal adapters are okay when an
> : async serial port is all you've got (or you're low on cash), but you'll
> : get better performance and more features with a real router every time.
> 
> The P25 is an old entry into Ascend's line, without the nifty SNMP, with the
> ability only to route one or the other of IPX and IP, and with less
> bandwidth (it sometimes can't handle a well compressed data load). 

I won't disagree with you on any statement about the "puny-ness" of the
P25, and would never recommend it to anyone who wanted a serious router
(along with the reasons you give, you'll also never be able to get
firewalls or IPSec encryption on it). However, I haven't heard any
rumors of it being discontinued (nor have I heard any rumors that it
*isn't* being discontinued).

> Now, if you want to know about the down-sides to Ascend's products,
> I'll be glad to forward you a just-shy-of-flame post I sent to the
> ascend-users mailing list about them leaving old product customers out
> in the cold.

Mostly, I enjoy it when Ascend users send mail to Ascend technical
support (or their favorite salesperson) complaining about the lack of
features that I'd also like to see. As an insider, I can talk about them
'till I'm blue in the face and it won't do nearly as much good as a
customer (especially a large or otherwise influential customer) putting
in a request for something.

If you'd like a list of some features I'd like to see (for regurgitating
back to proper channels in Ascend), I'd be happy to send them to you ;-)

As far as old product customers, it is bothersome when they can no
longer get fixes to obvious *bugs* in their boxes, but since our part of
the company is only working on new features for certain boxes, this is
something I don't have direct contact with so I can't comment on it in
any manner. Getting new features is often problematic, since new
features take more memory (RAM and flash) and often more CPU, and there
are certain limits that just can't be overcome (eg - see my comment
about firewalls and IPSec on the P25).