See comments at the end.
on 14/12/01 3:17 pm, Dan Grossman at dan@dma.isg.mot.com wrote:
>> snip/snip>
>>>>
>>>>
>>>> ---------
>>>>
>>>> In the section "Bandwidth on Demand (BoD) Subnets"
>>>>
>>>> The ID says:
>>>> "The resource shared may be a terrestrial wireless hop, a
>>>> satellite uplink, or an end-to-end satellite channel."
>>>>
>>>> These techniques are also used in other places than those listed. This
>>>> statement should be more inclusive. I suggest, for example:
>>>>
>>>> "Examples of the resource that may be shared include a terrestrial
>>>> wireless hop, a cable modem uplink, a satellite uplink, and an
>>>> end-to-end satellite channel."
>>>
>>> pen
>>> Well, if we define BoD that broadly, then there are number of places in
>> the
>>> section that need to be rewritten to take into account cable and
>> terrestrial
>>> wireless. Rereading the section, it appears that the principle holds for
>> these
>>> thing, although since propagation delays are not as large, the performance
>>> issues are not as great. That being the case, a little wordsmithing would
>>> generalize.
>>>
>>> Incidentally, terminology check: a "data over cable television upstream
>>> channel."
>>>
>> If we really want to extend BoD to all potential channels then even the
>> wireless LANs would apply (look at 802.11e; you could define BoD over it).
>> Do we want to just say the the technique applies to any shared media that
>> has a framing structure?
>>
> The common feature in question is not a framing structure (it does not exist
> for TDMA, for example), but rather a contention/grant/reservation scheme.
>
> If we accept this as the defining feature of BoD, 802.11 in PCF mode (which I
> believe is the basis for 802.11e, although I haven't been following it
> closely) is not a bandwidth-on-demand scheme, since it uses polling.
>
> BTW, the term "Bandwidth-on-Demand is badly overloaded. Can we come up with a
> better one? Perhaps Contention/Reservation would work.
>
> Also, the draft is pretty negative toward BoD. I hope that Mark will tone
> that down a bit.
>> Marie-Jose
> Dan
>>
We do need to get the definition of the term correct. I'm not sure this is
the same set of issues as in ASYM though where we didn't particularly
differentiate between types of reservation.
Can I suggest two extremes?
Reservation - where a fixed capacity is assigned for a fixed (possibly much
less than a session) time period. This is a rate-based assignment, and can
look very much like a "circuit switched subnetwork" to the IP traffic.
BoD - To me, I tend to think of the assigned capacity varies on a time frame
of the order of the link RTT. Capacity may be based on measurement of
the current transmit queue size and/or arrival rate and/or a traffic
predictor. The short-term, or *DYNAMIC* element is the key bit.
However, I accept there are a variety of uses of the term, and real systems
are often hybrids of various techniques, so we *MUST* be careful of wording
here so that the "audience" of the RFC understands.
Given what you are saying, I suggest we drop the DAMA and TDMA references,
or at least make this an example.
How do the 3GPP people understand the term BoD?
I like "BoD" as a section heading, because it's widely used, but it is
imprecise. It may encourage people to read the section!
Gorry
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:29 EST