NAT44 fixation in the "cloud native" world, and what to do about it [Re: Self-service tooling requires fine-grained authz -- it's NOT about the application protocol (was Re: (internal) DNS dysfunction is enterprise settings)]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Michael Richardson writes:
> Tony Finch <dot@xxxxxxxx> wrote:
>> This is outside the areas I am familiar with, but there's some really
>> interesting DNS stuff going on in the Kubernetes world, for dynamic
>> service discovery. But I get the impression they are using DNS for
>> compatibility with existing software, not because they care about
>> connecting their container cluster to a global namespace.

> Yes, but it's terribly *IPv4* NAT44-everything-you-can centric.
> It's nice that docker and kubernetes have made themselves IPAMs, but it turns
> out that turning that off is actually rather difficult.

This pains me as well.  There has been some effort to add IPv6 support
to Kubernetes.  Some of that was actually released in Kubernetes 1.9
(maked as "alpha"), but there's still quite a bit missing.  In
particular, the Kubernetes data model still assumes each pod has "an"
(one) IP address.  At the current level of IPv6 support (as far as I
understand it), this can now be an IPv6 address *instead of* an IPv4
address.  So there's no real support for dual stack yet; this is tracked
under

https://github.com/kubernetes/enhancements/issues/563

The efforts seem to have stalled a bit lately, so help in this area
might be welcome.  I'm not sure how to best engage with the vibrant
Kubernetes/cloud native community, but for all anyone who'd like to help
fix this and willing to hack some Golang => OPPORTUNITY

https://github.com/kubernetes/enhancements/issues/508#issuecomment-454838090
-- 
Simon.




[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Mhonarc]     [Fedora Users]

  Powered by Linux