Re: [babel] About BIER in Babel

On Tue, Mar 28, 2017 at 4:06 PM, Alia Atlas <akatlas@gmail.com> wrote:
> Hi Dave,
>
> Take a look at
> https://datatracker.ietf.org/doc/html/draft-ietf-bier-architecture
> and
> https://datatracker.ietf.org/doc/html/draft-ietf-bier-mpls-encapsulation.
> The latter allows use over Ethernet (despite the old draft title).  The IPv6
> encapsulation does not have WG agreement to adopt and wasn't discussed this
> IETF.

Even after reading more, I don't really "get it".  The list of use cases

https://datatracker.ietf.org/doc/html/draft-ietf-bier-use-cases

do not include most of babel's existing ones - I don't imagine more
than 5 tvs per household carrying the superbowl, nor a home with
thousands of cameras and hard disks streaming. At a data-center or
city-scale, I can see how some of this technology would be helpful
(but it's still difficult for me to wrap my head around using up 4096
bits in a single message)

Furthermore, babel is a dv protocol and doesn't need to carry all that
many bits, all the time. One bit pointing at the multicast superbowl
group within the local network would be enough.

I will read on, and perhaps enlightenment will strike.

> Regards,
> Alia
>
> On Tue, Mar 28, 2017 at 5:48 PM, Dave Taht <dave.taht@gmail.com> wrote:
>>
>> ... up until this moment I was mostly ignorant of bier. I may well
>> still be overly ignorant, but read on.
>>
>> On Tue, Mar 28, 2017 at 1:47 PM, Juliusz Chroboczek <jch@irif.fr> wrote:
>> > Dear Sandy, dear all,
>> >
>> > I think there are very good reasons why it would be interesting to
>> > experiment with BIER and Babel.  As you explained, there are at least
>> > two
>> > approaches worth considering:
>> >
>> >   - announce BIER routing directly in Babel, as your draft suggests;
>> >   - use MLD with Babel.
>> >
>> > I am not competent to judge which approach is suitable, but it is
>> > clearly
>> > attractive to run a single protocol.
>> >
>> > As I mentioned at the mike, there are a number of problems with your
>> > encoding.  My personal intuition would be to use a new AE number for
>> > BIER,
>> > but of course someone would need to write it down and check the details.
>>
>> However, leveraging this encoding over the others in the core draft seems
>> straightforward to encapsulate into an ae with no afteraffects
>> elsewhere in babel.
>>
>> https://datatracker.ietf.org/doc/html/draft-pfister-bier-over-ipv6
>>
>> 4096 bits (maximum requirement for bier), on the other hand, breaks
>> everything (babel mtu requirement while still fitting in everything
>> else),
>>
>> while (the minimum requirement) of 256 bits may well look a lot like
>> how the new source specific ae is formed.
>>
>> if there are other docs in that wg worth reading, running code on
>> linux, or other use cases, please point me at em!
>>
>> > As David mentioned, some of us would feel much more comfortable if there
>> > were a prototype implementation before we are asked to adopt your draft.
>> > I would be interested in helping, but I have neither the competence nor
>> > the manpower to do the implementation on my own.
>> >
>> > -- Juliusz
>> >
>> > _______________________________________________
>> > babel mailing list
>> > babel@ietf.org
>> > https://www.ietf.org/mailman/listinfo/babel
>>
>>
>>
>> --
>> Dave Täht
>> Let's go make home routers and wifi faster! With better software!
>> http://blog.cerowrt.org
>>
>> _______________________________________________
>> babel mailing list
>> babel@ietf.org
>> https://www.ietf.org/mailman/listinfo/babel
>
>

-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org