[195129] in North American Network Operators' Group
Re: IPv6 traffic percentages?
daemon@ATHENA.MIT.EDU (Jima)
Fri Jun 23 17:58:35 2017
X-Original-To: nanog@nanog.org
From: Jima <nanog@jima.us>
To: nanog@nanog.org
Date: Fri, 23 Jun 2017 15:58:21 -0600
In-Reply-To: <D572A249.7D84A%lee@asgard.org>
Errors-To: nanog-bounces@nanog.org
On 2017-06-23 09:09, Lee Howard wrote:
> But I think you’re asking for a business education series that goes:
> 1. Enterprise business consideration of IPv6
> a. It’s already on your network. All computers, tablets and phones have
> at least Link Local, and some set up tunnels. Plus, if your employees have
> dual-stack at home but single—stack VPN, you may not like your split
> tunnel.
Speaking with my Enterprise Day Job hat on...
I started doing analytics on IPv6 client support regarding a (currently
IPv4-only) employee-facing web app my team hosts. Turns out that in 15%
of connections with IPv6, the IPv4 is coming from a subsidiary VPN, but
the IPv6 isn't VPN'd -- supporting your point. (That number may be
artificially low, in that not all of our business units restrict the
users' source IP.)
> d. IPv4 runout doesn’t matter much to most enterprises. They only need
> a couple of addresses for new branch offices. Those enterprises who have
> their own IPv4 address block (from RIR, not ISP/LIR) should consider how
> much they could sell it for. At $15/address, a /16 approaches US$1
> million, which is real money to most CTOs.
When you get big enough (and go through enough mergers & acquisitions),
RFC1918 runout becomes a serious, legitimate concern. That's been a big
selling point for me.
Jima
