Re: [RRG] Thoughts on the RRG/Routing Space Problem

[Robin Whittle <rw@firstpr.com.au>]

I am replying to Scott, Iljitsch and Jari.

The MAB, UAB and micronet terminology I use here is discussed in:
  http://psg.com/lists/rrg/2007/msg00533.html
  http://psg.com/lists/rrg/2007/msg00535.html

Scott Brim wrote:

> While core routing/forwarding is untouched by granularity, there
>  will be more prefix mappings in ITRs.  There are several things
>  that mitigate that.  First, if there is any "pull" at all in the
>  mapping system, an ITR only has mappings it uses, so small sites
>  that only connect to a few places can have small ITRs.  Second, 
> an ITR need not cache everything if there is a mapping node 
> nearby that will do the caching for it.  Third ... well, I just 
> woke up with melatonin in my brain, so it isn't quite working 
> 100% yet, but there are mitigation techniques in all the various 
> schemes.

Another mitigation technique is a full database (push) ITR reducing
the cost of handling every mapping update in the world by updating
its RIB with only those updates which are required by the current
traffic.  This is towards the end of my message:

  Re: [RRG] Thoughts on the RRG/Routing Space Problem
            - millions of micronets
  http://psg.com/lists/rrg/2007/msg00684.html


Iljitsch wrote:

> However, I disagree that it's necessary or even useful to make
> a syntactical distinction between existing PI space, the new
> type of PI space and locator space. Anything that's in the BGP
> tables is either existing PI space or locator space. Which of
> the two isn't terribly interesing. Anything that's not in the
> BGP tables is either unused or new PI space so it can safely
> be routed to an encapsulation device.

I agree.  I think there's no need to distinguish between the use of
address space, or between types of BGP advertisement, based on
whether the space is used in a conventional manner or as an MAB.

A MAB is advertised by many ITRs - I think "anycast" is an
appropriate term for this:

   http://psg.com/lists/rrg/2007/msg00683.html

But that doesn't require any new distinction beyond the fact that
there would be some list or database in the ITR-ETR system listing
all the MABs.  It is possible to make the system work with a single
ITR advertising the MAB in the DFZ - but it is more efficient in
terms of load spreading and optimal path lengths to have two or more
likely thousands, of ITRs advertising it.


Responding to Iljitsch, Jari Arkko wrote:

> The crux of the issue is whether there's any form of pushback for
> users and organizations to insert their identifier space into
> the global tables. If there is none, and there is no incentive
> that makes users avoid that, the chances are that no amount of 
> additional protocols will actually solve the fundamental problems
> we have today.

By "identifier space" I assume you mean the space which includes
host addresses, with this being either traditional BGP-managed PI
space - or the new kind of ITR-ETR managed "micronet" space, which
does not involve any new route being advertised in the global (BGP)
tables.

A form of pushback would be that the traditional PI space winds up
being significantly more expensive for many end-users than the new
type of ITR-ETR managed "micronet" address space.

The idea with Ivip is that each Mapped Address Block (MAB) - which
requires a BGP advertisement - covers the micronets of many
end-users.  As long as this is the case, the system is greatly
reducing the number of BGP advertisements compared to what would
happen without such a system.

The same is true of LISP Proxy Tunnel Routers:

  http://www.1-4-5.net/~dmm/draft-lewis-lisp-interworking-00.txt

  5.1.  PTR EID announcements

     A critical characteristic of PTR functionality is to
     advertise aggressively aggregated EID-prefixes.

I think this means the same thing - as long as there are lots of
micronets (EIDs) in the MAB (prefix handled by PTRs), then the
overall goal of reducing BGP advertisements is achieved.

Although I think there needs to be a charge for every change in a
micronet's mapping, as I discussed in the above-mentioned message, I
think that having a micronet - or a UAB (User Address Block) within
the MAB which each end-user can split into micronets as they wish -
will be a lot cheaper than having a PI prefix today.

The end-user only needs to gain their address space with whoever
administers the MAB.  So the end-user doesn't need to known anything
about advertising BGP prefixes and they don't need to deal with an
RIR, unless the MAB is run by an RIR.

Assuming the address space managed by the ITR-ETR scheme doesn't
suck (in terms of path length, packet losses, PMTUD problems etc.)
and that there are lots of end-users who are happy with:

1 - Either less than 256 addresses (I am assuming IPv4) - which is
    bound to cost less than a PI /24 - or more addresses, but they
    are happier with ITR-ETR managed micronet space than ordinary
    PI space, because it still winds up cheaper.

and

2 - They don't need the space to run ETRs or a few other things
    which ITR-ETR-mapped address space is unsuitable for.

then I would expect this to result in "pushback" against them
getting ordinary PI space.  In addition to the new type of address
space being less expensive, there will be other advantages, or at
least not disadvantages:

1 - The new ITR-ETR managed micronet space will be as portable
    as ordinary BGP managed PI space - as long as the ISPs run ETRs.

2 - Depending on the characteristics of the ITR-ETR scheme, this new
    type of address space may have superior performance to ordinary
    BGP managed PI space in terms of flexibility, speed and
    reliability of TE and multihoming service restoration.

Furthermore, at the RIR level, micronets will enable a much more
efficient use of address space, in terms of the number of end-user
networks served per large address block - since each end-user can be
given whatever space they need without the usual 256 IP address
granularity of the current BGP system.

So RIRs might want to encourage the use of micronet space by lower
fees and/or making space more available to ISPs (or whoever) is
running the MABs.

All this should result in "pushback" - traditional PI space being
less attractive and/or more expensive than micronets - for a large
number of end-users.

 - Robin




--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg