Re: [Idr] draft-decraene-idr-reserved-extended-communities-00

[<[e-mail]>]

> > Reserving 2^48 = 2.8 10^14 = two hundreds thousand billions
communities
> > while a single is required, I call this wasting resources.
> 
> The resource being wasted is the 2^48 potential values of the low
order six
> bytes of the extended community??  I.e., there are 2^48 values that
could
> have been used for codepoints but are instead going wasted?
> 
> How many codepoint values do you expect IANA to assign from this
registry?
> Even if they assign, say, a billion, the resource utilization will
still be
> pretty close to zero.

True. (I could argue that 8 octets are still reserved and could be
further used, but I don't think it's really the main point)
Yet under your hypothesis, it would be much better than the alternative
(using a billion community types).

> > We ran out of class B, AS number, IP addresses, optional parameter
> > length... How can you be certain that we will never run out of
extended
> > types?
> 
> I don't think the proposed registry is a very good hedge against
running out
> of extended types, as it essentially creates an extended community
that
> consists of eight bytes of type field and zero bytes of value field.
If we
> use up all the two-byte extended type values, we will probably still
want
> some value field bytes.

It's a small contribution. It's effective in the case where more people
want to get a non transitive community. (or a block of communities).
Given that the cost of this proposition is very low, it seems to me that
potential gains are greater than the cost.
 
> > if _you_ believe these are evil, you will be able to remove or not
> > interpret them.
> 
> It's hard to imagine that all 2^48 codepoints will be evil.
> 
> > A registry for reserved/well known/std action community do exist. Do
you
> > have a concern with it?
> 
> Yes, I don't think it makes any sense to have a registry for
communities
> that "have global significance and their operations shall be
implemented in
> any community-attribute-aware BGP speaker"; obviously one can't
allocate a
> new codepoint from that registry without throwing all existing
> implementations out of compliance.

- You are quoting RFC 1997 (BGP Communities Attribute) which has defined
such "Well-known Communities". Discussion on
draft-decraene-idr-reserved-extended-communities-00 will not improve
your concern on RFC 1997 well-known community. (And BTW this registry is
not anymore about "Well-Known Communities", but about "reserved BGP
communities" so there is no such concern
http://www.ietf.org/mail-archive/web/idr/current/msg03002.html  ) 

 
- We are _not_ proposing to create a registry for communities that "have
global significance and their operations shall be implemented in any
community-attribute-aware BGP speaker". We are proposing to create a
registry to allocate values which are globally unique.