Server addressing for renumbering ease

Mark Smith nanog at 85d5b20a518b8f6864949bd940457dc124746ddc.nosense.org
Mon Nov 8 09:27:58 CET 2010


On Sun, 7 Nov 2010 15:56:25 -0500
Ben Jencks <ben at bjencks.net> wrote:

> On Sun, Nov 7, 2010 at 15:22, Mohacsi Janos <mohacsi at niif.hu> wrote:
> > On Sat, 6 Nov 2010, Ben Jencks wrote:
> >
> >> According to [1], they used EUI-64 addressing on their servers so that
> >> they could renumber easily. The common objection to that is that you
> >> wouldn't want to change DNS anytime you swapped a NIC (and thus a mac
> >> address). The obvious alternative is static addressing, but that makes
> >> renumbering a pain. There's also DDNS, but that seems like a bigger
> >> headache than either of the above [2]. What are people here doing?
> >
> > Using autoconfigured addresses on servers are bad idea. If you want to
> > generate big problems for yourselves, use it on DNS servers....
> 
> For DNS servers it would obviously make things difficult, but there
> are massively fewer DNS servers than other servers. Touching every DNS
> server during a renumbering seems fairly reasonable; touching every
> server can be infeasible.
> 
> Why do you find it a bad idea for other servers? Just the switching
> NICs/mac address dependence issue? I'll grant that SLAAC is another
> moving part, but it's a pretty simple protocol that I'd be willing to
> trust with numbering my servers.
> 
> Another thought I had is that with the rise of virtualization, SLAAC
> actually starts to make more sense for virtual machines.

It's more than makes sense, it is really quite convenient. I have a
virtual host that when I boot it I bridge it out onto the
network. It then picks up a prefix from an RA, performs SLAAC,
and then makes available it's hostname and SLAAC address via multicast
DNS.  I then just 'ssh <hostname>.local' to access it, never knowing or
having to care about what it's IPv6 address is. It reminds me of
accessing Novell servers in the 90s ...

> While
> replacing a NIC in a physical server isn't uncommon, since VMs have
> virtual mac addresses they can reasonably be expected to keep the same
> mac, and thus the sam SLAAC IP, for the lifetime of the VM.
> 
> -Ben
> 
> > In 6diss/6deploy training material we are advocating using static addresses
> > for servers:
> > http://www.6deploy.eu/tutorials/131_IPv6_deployment%20consideration_v0_8.pdf


More information about the ipv6-ops mailing list