Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: (ASCEND) POTS bonding
At 10:31 AM 5/7/98 -0400, Phillip Vandry wrote:
>> If you are scared of ISDN you could connect using two modems (56K ea)
>> and get a bonded rate of 112K. I've heard this works between non-similar
>> devices also, like one ISDN and one POTS modem... But I would imagine
>> that would cause problems as frames outgoing on the ISDN would arrive
>> before the frames on the POTS line.
>
>The last time I read the spec (RFC1717, I think there's a superseding RFC
>now), supporting incongruous links in the same bundle was left as a
>problem for the implementor. Implementors are encouraged to use a bit of
>intelligence, such as by sending two thirds of the packets out one link
>instead of only half if that link is twice the speed of the other one.
>
>But I have a feeling that, given the lack of practical applications and
>the extra complexity, such intelligence has not been widely implemented.
>
>How does the Max handle it?
The MAX will most likely be on the "answering" side of any "weird" combination
of link types. In the incoming direction, obviously the remote client drives
where the packets go, and if they are fragmented or not - the MAX will handle
either way. On the outgoing side, the MAX will send to the next 'free'
channel.
The MAX does not send fragmented, but whole frames independent of the link
speed. If one frame is being sent over a slow speed link and then two other
frames are destined for the same endpoint, it could well send them both
over the faster link - depending on if the first frame was completed or
not....does that make sense?
Kevin
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>
Follow-Ups:
References: