Merit Network
Can't find what you're looking for? Search the Mail Archives.
  About Merit   Services   Network   Resources & Support   Network Research   News   Events   Home

Discussion Communities: Merit Network Email List Archives

North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: DNS DOS increasing?

  • From: E.B. Dreger
  • Date: Mon Jan 21 21:01:22 2002

> Date: Mon, 21 Jan 2002 18:50:07 -0500 (EST)
> From: Stephen Griffin <stephen.griffin@rcn.com>

> With the added benefit of not increasing the routing table
> size. The downfall is (potentially) increased address
> consumption (perhaps mitigated in that small entities doubling
> a /28 is still better than lying cheating and stealing for a
> "large" PI or PA block).



Let's pretend for a moment that all IPv4 space is unallocated.
Clean slate.

We have room for ~ 2^16 ASNs and ~ 0.7*2^32 addresses.  This
works out to a bit under a /16 per ASN on average, assuming that
we use all ASNs.  Considering that we're running out of IP space
faster than ASNs (even after discounting wasteful 4/8-, 12/8-,
38/8-, etc.-style allocations), I think that this is
conservative.

Let's also carve up IP space similar to the original classful
method, specifying a "natural" granularity... say /10 for 1/8
through 8/8, on to /28 for 223/8.

If SMD can singelhandedly (or close to it) define the top of
swamp space, surely a group can similarly devise an allocation
policy and filtering BCP. ;-)



When one needs IP space, one receives a "natural" size subnet
with room for growth.  e.g., perhaps I would get 223.0.0.0/28,
with the rest of 223.0.0.0/26 reserved to "grow into".

If I'm growing slowly enough, then 223.0.0.32/27 might be
allocated to someone else.  If I need no more space, perhaps even
223.0.0.16/28 will belong to another.



Contiguous blocks are friendlier on the routing table.  Imagine
for a moment that each ASN needed one and only one subnet.  Yes,
I know that this means coarser granularity and more waste... but
two or three subnets would give more flexibility and still yield
a far kinder, gentler table.  (When one hits, say, five subnets,
perhaps one would have to renumber two or three into a shorter
prefix within a year.)

Granularity grouping means that providers can filter based on
subnet length without anyone getting upset.  I'd not be so
bothered by Verio-style filtering if it didn't plonk so many
IMHEDO[1]-valid routes.

[1] In My Humble Edge-Dwelling Opinion



I think that we can have conservative IP allocation and
reasonable routing tables if space is administered differently.
People will attempt to hoard any valuable resource... less
valuable, less hoarding.  Better auditing, less hoarding.  You
want new space, you justify your old -- including legacy /8
blocks.



Yes, I know that this is all a bit idealistic.  But I don't think
it's so idealistic that none can creep into reality.  (Then
again, look at the rate at which swamp space has been returned
since the introduction of CIDR...)



Eddy

---------------------------------------------------------------------------
Brotsman & Dreger, Inc. - EverQuick Internet Division
Phone: +1 (316) 794-8922 Wichita/(Inter)national
Phone: +1 (785) 865-5885 Lawrence
---------------------------------------------------------------------------

Date: Mon, 21 May 2001 11:23:58 +0000 (GMT)
From: A Trap <blacklist@brics.com>
To: blacklist@brics.com
Subject: Please ignore this portion of my mail signature.

These last few lines are a trap for address-harvesting spambots.  Do NOT
send mail to <blacklist@brics.com>, or you are likely to be blocked.





Discussion Communities


About Merit | Services | Network | Resources & Support | Network Research
News | Events | Contact | Site Map | Merit Network Home


Merit Network, Inc.