Re: [SAGE] Naming conventions for servers, network gear, etc.

["Kurt Buff" <kurt.buff@gmail.com>]

On 1/6/07, Etaoin Shrdlu <shrdlu@deaddrop.org> wrote:
> Jason Antman wrote:
>
> >>Have there been any papers done on naming conventions for servers,
> >>network gear, etc?
> >>
> >>
> >I work for Rutgers University, which has pretty large IT operations
> >department. University-wide, the hosts are named using department-specific
> >names, which are a hodgepodge of functional names  such as NBCS for New
> >Brunswick Computing Services, and "random names" - Eden, Clam, etc.
> >Clustered servers are named such as Eden-u1, Eden-u2, etc.
> >
> >
> I have found over time (and I've seen a lot of time) that *any*
> consistent naming scheme works. It's helpful if the names have some sort
> of internal sense that makes NBCS-Room449 something that is part of your
> responsibility, where Eden-u1 is clearly someone else's (just for
> example). Another thing that's important is the long view. As we know,
> certain operating systems and applications do *not* like being renamed
> (think Oracle; think Windows2k3). The less emotional attachment there is
> to a naming scheme for servers and network devices, the better.
>
> Let me reiterate the excellent work already quoted so many times. I
> don't think it gets better than that.
>
> http://www.nanog.org/mtg-0405/ringel.html
>
> >For my personal networks, I really prefer names that have nothing to do
> >with the functional nature of the machine, as I view this as making
> >network reconnaissance too easy. My home development network has "SATURN"
> >as the main DHCP/DNS/LDAP server, and the other machines are named after
> >Saturn's moons, allowing approximately 56 unique names.
> >
> >
> Amusing, but what on earth makes you think anyone interested in your
> personal network will care in the slightest about what the *name* of
> anything is? If it's interesting enough for someone to care, they will
> *know* that you have all those services running on just one machine, and
> it'll be over. OVER. On the other hand, unless you have the ability to
> support a bunch of bots, I doubt that the bad guys care. Everything's so
> automated now, anyway.
>
> My router is usually named rooter, because it makes me laugh. When it
> comes to small networks, I'd just as soon have names that are memorable
> for the entertainment value, as anything. Some of my favorites have been
> muscle cars (Cobra, Mustang), Nobel Prize winners for physics (with an
> honorary mention for Hawking, who *should* win), and insect predators
> (Mantis, Widow).

Heh. When I was working in a small startup some years ago, I decided
that various breeds of chile pepper sounded yummy for my machines -
habanero, jalapeno, ancho, piquin, etc. When I got to a larger site
(more than about 12 servers), I started to name them after functions
again. And made the names as short as possible - fs1 for the first
fileserver, etc. - because it simply made more sense. I've never had
responsibility for more than 50 servers at a whack, but I'm now in a
position where we have over 400 PCs on the net, and I'm enforcing a
stardard of the 2-or-3 letter department code, a dash and the
username. Haven't had any collisions in that namespace yet, but if we
do, I'll just start using middle initials, or append a numeral, or
something like that.

If I ever become responsible for a larger environment, then the ideas
in the Ringel presentation will definitely prove useful.



Kurt