[194522] in North American Network Operators' Group
Static IP allocation schemes for end users (commercial)
daemon@ATHENA.MIT.EDU (Graham Johnston)
Fri May 5 10:57:47 2017
X-Original-To: nanog@nanog.org
From: Graham Johnston <johnstong@westmancom.com>
To: "'nanog@nanog.org'" <nanog@nanog.org>
Date: Fri, 5 May 2017 14:57:41 +0000
Errors-To: nanog-bounces@nanog.org
I work for a cable MSO, meaning that our access network is DOCSIS based. 15=
years ago when we had way more IP addresses than customers we had a static=
IP allocation scheme wherein we aligned a /24 with each node and reserved =
the first 20 or so IPs for static assignment, the rest being left for dynam=
ic assignment. The primary reason behind this scheme was that as the networ=
k grew and a node moved from one CMTS to another we could pull the /24 with=
it and customers wouldn't have to re-address.
As our network has grown, in terms of the number of nodes, as well as the n=
umber of IPs in use we've come to the obvious conclusion that the old schem=
e isn't workable into the future. For instance I will soon have more nodes =
in service than I have /24s allocated to me.
Instead we are entertaining two basic options as an alternative.
* Create static reservations, as required, within the otherwise dynamic=
range.
* Whether or not the customer continues to DHCP or assigns the IP st=
atically doesn't matter to me.
* Move to having a single subnet, or portion therein, per CMTS. This ke=
eps the process more or less identical to what we do now.
Both of the schemes above would require the customer to undergo addressing =
changes in the event that we move their node between CMTSs in the future.
Can anyone else share what they are doing or otherwise identify if there is=
a best practice in this area?
Thanks,
Graham
