AWS Elastic IP architecture

Mon Jun 1 22:41:19 UTC 2015

>The question that Matthew Kaufman proposed was specifically asking about 
>app architecture deployments, so what Facebook is choosing to do is 
>entirely germane.

I'd lean more on the "ipv6 evangelism" side of the discussion, but:

Facebook controls the whole stack and can require buy-in from their apps 
people to push IPv6 first.  If that costs them dev time to patch some OSS 
to handle it gracefully, that's their business decision.  In the "cloud 
host" domain, you're dealing with a much more heterogeneous environment 
where the provider doesn't control the whole stack up to the apps.  Making 
the platform as frictionless as possible for customers is key; customer X 
is not going like your platform much if widget Y doesn't run properly 
"because IPv6".  Sure, widget Y should get its excrement together and 
handle it, but all customer X sees is "widget Y fails on provider A, but 
runs fine on provider B" where provider A was v6-only internal but provider 
B is either v4-only or dual stack.  Guess where customer X spends their 
dollars now?

I'm on your side, here: I run my own stuff v6-first wherever possible and 
have filed bug reports, submitted workarounds/patches, etc.  We need people 
doing that to push things forward.

On this given point, though: Facebook -ne generic hosting platform

-- 
Hugo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20150601/ae8781f2/attachment.sig>