Re: [v6ops] Operational Consensus on deployment

[Owen DeLong <[e-mail]>]

On Aug 5, 2014, at 5:56 AM, William F. Maton Sotomayor <[e-mail]> wrote:

> On Tue, 5 Aug 2014, Tore Anderson wrote:
> 
>> * Brian E Carpenter
>> 
>>>> (In parenthesis, I've never seen sunsetting IPv4 as a real problem.
>>>> One day somebody will notice that there are no more IPv4 packets. But
>>>> that is many years in the future.)
>> 
>> The thing is, if you've built your infrastructure on IPv4, removing it
>> is going to be a royal pain in the arse. Even if all the IPv4 users have
>> left the public internet, if your application server speaks to your
>> database server over IPv4, and the database server speaks to the iSCSI
>> array over IPv4, and so on and so on, then you're simply not in a
>> position to easily shut IPv4 off.
> 
> Worse still is all the gear to workaround NAT, so it isn't just the usual suspects of desktops vs servers vs appilcations, but also the point-to-point setups (video conferecing comes to mind) I have encountered that require going around NATs.

Once the video conferencing setup is running over IPv6, will it care if you remove the IPv4 NAT workaround infrastructure?

Perhaps I am confused.

> 
> (There is something to be said about a net gain in maintenace overhead in elminating some of that, but there's also the effort to remove it and test, etc.)

As a general rule, once nothing is depending on something, removing it is usually fairly quick and painless, often accompanied with much celebration and the consumption of adult beverages with bubbles in the case of perpetually painful legacy infrastructure like NAT gateways and NAT workarounds.

The hard part is getting all that other stuff that depends on such infrastructure to the point that it depended on such infrastructure (past tense), but no longer does.

Owen