Slashdot Mirror
Suitable Naming Conventions For Workstations?
spectre_240sx writes "We've discussed server naming a fair amount in the past, but I haven't seen much about workstations. Where I currently work, we embed a lot of information in our workstation names: site, warranty end date, machine type, etc. I'm of the opinion that this is too much information to overload in the machine name when it can more suitably be stored in the computer description. I'd love to hear how others are naming their workstations and some pros and cons for different naming schemes. Should computers be logically tied to the person that they're currently assigned to, or does that just cause unnecessary work when a machine changes hands? Do the management tools in use make a difference in how workstations are named?"
And that's saying something.
Honestly, can you even think of a stupider question? How is this even an issue? Just name each machine with an ID and put the information in a spreadsheet somewhere. It's not a complicated problem.
don't name by person just makes it harder to do swaps, moves, and other stuff. Also times you need a open system that many people uses. warranty end date, machine type + where it (general area) is seems good.
zombie-pron-server-1 ... ... ...
zombie-pron-server-2
zombie-pron-server-3
zombie-pron-server-4
zombie-pron-server-5
Like an ID for a database record, the name should be unique, mean nothing out of context, and used only to look up a description of all the information you are trying to encode in it. What happens if the warranty info changes? What happens if you assign the wrong machine, move where it is located, or change some other fungible property (either through upgrades, or simply because you encoded the wrong info?). You don't want to have to go through machine renaming exercises, updating dns entries, etc. or have to live with the degredation of your naming convention.
How about
MicrosoftSpamBot01 thru MicroSoftSpamBotxx?
Name them after Star Trek ships, races, planets and character names. You are obviously not a true CIS geek.
Better known as 318230.
I've seen a variety of things done. Personally, I named computers by division and assigned the computers from there. This worked fine for a group of about 25 people, but could be problematic when running into larger groups. I'm not sure how a larger group would do it, but I'm sure it would be done somewhat differently.
Don't forget that high turnover could make administrating the PC's a mess.
A computer name should not be a database. If you want to store information such as site, warranty end date, machine type, ... use a database.
This is the dumbest Ask Slashdot, evar.
I want to delete my account but Slashdot doesn't allow it.
Well, I typically rebuild an entire system when they are reassigned so tying the name to the user is possible but I still try not to as it ends up cluttering my AD. I prefer standardized names which tend to include the Site, Department, Type (laptop/desktop), and a number identifier so a customer in Boston MA in the Accounting department with a Desktop might be BOSACCDSK0001. Very bland and unimaginative but it works well.
Give each station a unique id for its name, and store all the other information in a spreadsheet.
He insisted that all names came from Alice in Wonderland. Very annoying. And not practical.
I name them by location eg. Building-Room-{Front, Back, Left, Right...} makes tracking them down a bit easier provided no one moved it...
i.e. gla-hub-04a-001
or here's a off the wall idea...
Number them as: City(or location)+machines static IP address within the internal network.
i.e. Glasgow-10-10-11-124
simples....
Laters Sol "Have you found the secrets of the universe? Asked Zebade "I'm sure I left them here somewhere"
Make it short and simple, there is nothing worse than pc names like hdh02039024904. Also, refrain from using names like enterprise, chewbacca, kryten as they make you look like a real nerd.
Our desktop guys reimage workstations rather than relying on the last user to not fuck things up. Hostname contains department, user's name, and OS.
They're short, English (or English-like) and easy to keep in mind when you're using them. It beats the last place I worked where it was something like your logical purpose in the company (prog-, admin-, etc.) plus your username.
Put the extra information in a database or spreadsheet, not in the machine name.
My first workstation was named tangent (after myself!)
My second workstation was named sine, followed by cosine, secant, cosecant and cotangent.
I got stuck for a while before I decided to go with arctangent, arcsine, etc but that didn't last
So out came hyperbolictangent... and I promptly gave up and now I name them after hot young female movie stars.
Morale of the story: Make sure your naming convention has room for expansion.
I'm not a fan of crazy overloading(the name has to be unique in any case and I'd rather do a lookup if I really need the warranty details, rather than stare a nasty truncated version of them in the face every day); but what works best really depends on how computers are used in your organization.
For instance, if you have laptops, individually assigned to employees, and relatively low turnover, a name that tells you about the machine's primary user is really handy. It allows you to instantly associate the voice on the other end of the phone, or the name on the trouble ticket, with the machine in question.
If you have desktops, location based naming might be more useful, particularly if users move around, are replaced frequently, or share hardware per shift or something.
It's hard to give general rules for naming because, in essence, a name should capture(as succinctly as possible) the salient characteristics that make something unique. Exactly what those characteristics are depends heavily on how your organization is set up.
and keep all the other information (hardware type, user, etc) in a separate database using the name as the key. I worked in a large shop and everything (machine name, ip address, etc) all came from a server when the machine booted up. This made it easy when a machine failed. It only took a couple of updates at the server to assign ip addresses, names, users. etc if the hardware failed or was upgraded.
We used to name our machines by location (room number), but renovations and office swaps are far too common to do that anymore. We now name the machine by the username of the primary user. Since we have a policy of reimaging a machine whenever the user changes, this also acts as a reminder to us if it somehow was skipped. We also add to the end of the machine name an L or a D depending on if the machine is a portable or a desktop. True, we're smallish -- only about 100 PCs.
Now the old place I used to WORK, the machine names were all people-friendly FOUR letter words. PEAR, LEAP, HAZE, etc. This is because they were public terminals in a library and the printouts (at that TIME) came out by the machine name rather than user name. The BOSS always took GLEE in picking out a new WORD for the name on the rare occasions there were new terminals. In this case, the name was defined by a fixed location. We regularly swapped around machines and renamed according to the desk space's name.
Their names are "TootToot", "WootWoot" and "S3xB0vin3"
People move, machines get re-allocated, rebuilt, etc.
I use the service tag. Why? Several reasons:
Stuff like "bob-pc" or "accounts1" does not scale and either becomes inconsistent, or you need to keep renaming PCs which presents other issues (fucks up any configuration databases you have, etc).
So, service tag - boring as fuck, but does the job.
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
but... that's the same set...
ok, i didn't actually fuck any of the celebrities. but i don't fantasize about them that often either, so they don't really count for this purpose.
...but I'm a big fan of giving machines actual names, after TV shows, bands, movies, fiction, etc. I prefer to log into "Trixie.mycompany.com" instead of "LAUX001"; the former, in addition to being easier to remember, just gives the machine a trifle bit of "personality". Yes, I realize that the latter may convey more information (mail servers especially seem to do this: "CHIMAIL01", "NYCEXCH05", etc.), but it feels cold and impersonal; if you treat your machines as just machines, as just any old random tool you'd grab and work with, then they become just a series of interchangeable parts. Giving a machine a name invokes something, typically whimsical, that just adds a touch of humanity back into the system. Yes it's still a machine, yes it's going to spit out a thousand nonsensical errors when you forget a semicolon somewhere in your C++ file, and yes it will eventually be replaced, but for that period of time when you're working with it, you're just that little bit more connected to something more ... personal.
Maybe this is just old school thinking; it seems like this was much more common back when everyone had an account on the campus Unix boxen, complete with subtle importance ("Oh, you have an account on Kramden? That's a much faster Vax than Norton...what project are you working on that you scored that??").
We recently went through this where I work, one of the suggested ideas for this was this very crazy cryptic scheme that I found was difficult to read and self referential -- not very good when you are dealing with a limited number of characters. We ended up settling on [City or special campus]-[Department]-[Asset number]. We are on a smaller scale, mind you, but we should be able to easily deal with up to 10,000 machines on our network and with some reworking, many more. My main point here is that simplicity works the best. Make sure you keep an inventory database of equpiment and you should have no trouble as long as you stick with the convention.
imagine the horror of walking into a lab where all the workstations are named OMG-David-Caruso-01, ...
OMG-David-Caruso-02,
*shudders*
...after all the boring low power beige posters who think your question sucks.
You can use my name for the zooty new multi-core with the blue leds.
nz-aco-01
vm-somethingshorterthanwhatimtypinghere,maybedescriptive_iftheworkstationhasaparticularfunction.e.g.dev,reception,etc,etc-01
Oh and the number at the end can change. ie 02, 03.....99.
basic format of,
country-.....-#num
Hell you could even have something like
department-....-#num. Especially if your departments are particularly distinct where it would be best to reinstall the OS during a move. ( ie from a dev to a receptionist ).
...your machine was called "Duchess".
Keep it simple. I work at a college, and what we do for desktops is, we name them after location, room, number of workstation. So if the workstation is at our aviation campus in room Y109 and it's the 3rd workstation, it would be AVY10903 (AV-Aviation, Y109-Room, 03-3rd workstation) Laptops, we tie to users, we give it the users login name as the laptops name. We find this easy so when we have staff/faculty turn over, we are not running to workstations to rename them, and if its a laptop user that is being replaced, the laptop is returned to IT and we get it ready for the next user. This may or may not work for you, but it works for me.
I named one of our machines 'yomomma', one of my colleagues started fuming about the lack of professionalism and said we should name the machines something like 'lab102'...
After a while we where called into a meeting with the chair with the topic 'how to name the lab computers' ( I was the admin).
My office mate started suggesting the names of obscure tropical fish.
I think I might have suggested the names of the seven dwarfs...
After a few minutes of such political incorrectness, the chair stormed out of the meeting after a short rant.
I eventually gave in and changed the machine names to various
birds of prey...
I would not have thought that machine names could arouse such a passion.
There's no need to get fancy with your names. Anything beyond an ID tag will be outdated in no time. Users change, machines move between buildings/rooms, services change, etc.
We just use the simple company-id# for all workstations. When we opened another office in a different city, we tagged all of those as city-id# to distinguish. Wasn't necessary, but we wanted to be able to spot those at a glance.
In the end, you should have an Asset Management solution in place that will track all the extra cruft. Building/Room #, assigned user, warranty, purchase date, services running, etc. No need to stick all that in the name.
Now, for my sandbox network that gets wiped often to test new tech configurations, all machines are named after rain deities. We don't track those in the EAM suite, and they wipe so often that names don't even matter.
Dell service tag - because people who will get them (in this economy) are not going to last as long as your company's "workstation" leases.
Oh, place your computers and servers in the right containers in Active Directory, for easy management and policy assignment.
Why the hell do you need to give servers or any physical asset in a company names!!, it's not like they will come to you when you call out. The system we have were I work is simple, everything has a bar-code label with 2 number sequences with a space between them. The first sequence is 4 digits and designates the end of warranty period the rest is just a 6 digit sequence (numbers and characters). Simple and to the point, if you want to check who was the last person assigned that asset just look it up in the asset management database. As for names on the network for servers/workstations etc, yes we use the asset number as the machines name for login purposes etc.
On the other hand, it would be ideal workstation to create a GUI interface using Visual Basic to track an IP address.
We used to name our machines after Lovecraftian deities but some of the sysadmins got grumpy when they couldn't pronounce the name >
Easy. Name them after nebuli. My workstation is named NGC 6060, for example.
I name them all "Steve".
Shanghai
Mumbai
Buenos Aires
Moscow
Karachi
Delhi
Manila
Sao Paulo
Seoul
Istanbul
Jakarta
Mexico City
Lagos
Lima
Tokyo
New York City
Cairo
London
Tehran
Beijing
eventually you'll get to places like Holyhead, Waco, Palo Alto, Bakersfield, Piscataway, Sudbury, Guelph, Alice Springs, etc.
RS
Shoes for Industry. Shoes for the Dead.
Serial. Numbers.
Either assign them yourself (00001, 00002, 00003...) or use the manufacturer's serial number.
(Now, if you're doing it for your home network, that's a different story. Use the names of known ring-bearers, or secret identities of the Justice League, or actors who've played the Doctor, or starship captains, or whatever you find amusing.)
http://alternatives.rzero.com/
One of our telescopes had computers named after German beers, since it was installed by German engineers. The main control computer was called kronen. The other telescope was set up by a Tucson guy, so it uses Mexican beer names. The main fileserver is named corona, for instance.
The determined Real Programmer can write Fortran programs in any language.
How about these:
- abcdef
- ghiklm
- nopqrs
- uvwxyz
Then there is always:
- bob
- alsobob
- theotherbob
- notthatbob
- bobby
- bobbydoesdallas
- bobbob
Name them whatever you want, since chances are by the time you get enough computers you usually have someone who decides on boring names like:
- l00312
- l78302
Simply because it makes inventory easier. In the meantime decide amongst yourselves and choose something that you like. Remember to take into account how many names of the same theme you can come up with.
Jumpstart the tartan drive.
Great except for my credit rating.
WhatMeWorry!
I am a grad student in a physics department at a major university.
The grad students have access to a lot of machines around the building as workstations, and they're all named things like lagrange, maxwell, gauss, etc. (Bohr, newton, faraday, and the like are servers.)
Individual professors get to call theirs whatever they want -- my advisor's two are klingon and romulan.
If your company uses asset tags make the machine number the asset number. At least you'll be able to find it network wise, and when it get re-imaged the machine name will be easy to figure out no matter how hosed the original drive.
Also, if the machine changes users, the asset number is still relevant.
The preceding post was not a Slashvertisement.
I have a long running argument with some of my coworkers about names for software deliverables.
I insist on something you can pronounce and preferably something that makes sense and gives a strong indication what it is. If you are really desperate, call it something cutesy that people will at least remember.
They want to use incomprehensible, unpronouncable, random strings of characters One True Official Company Blessed "product codes".
I asked them for where the One True dictionary of product codes is. There isn't one.
Ok says I. I'll call the software deliverable that if I can look at the back of the hardware device and see that string of characters.
Nope. Can't.
So the three of them overrule me and I left them to it.
Much though I detest the army... ye olde british army storemans habit of general to specific naming is Good. "Trousers - Mens - Battle dress - Khaki - Large" at least allows a dumb troopie to search through a pile of trousers and sort them on to the right shelf.
(Hint: if you can't think of a good meaningful sentence describing what you are building... you probably shouldn't be. You are building a hodgepodge and a mess.)
Two hours later I came back and they were _still_ arguing about _which_ was the The One True Company Blessed Product Code.
Hokay says I. You have convinced me. You have convinced me that if you want to change the names of the software deliverables to these garbled bits of line noise that you guys can't even agree on... you'll have to do it yourself. I won't.
Last I looked my readable / understandable names still held.
At $LARGE_COMPANY where I work, most workstations get an IP address from a DHCP server (doesn't usually change, in the form of dhcp-ddd-ddd-ddd-ddd.example.com, where ddd = dotted quads of IP addresses and example.com is replaced with our actual domain. Since most of our staff have notebooks, this is useful for (among other things) figuring out where a machine is located. It also scales well.
For desktop workstations with a static IP (most of them are DHCP, but if you have a plausible reason why you need a static IP, you can get one), you can pick your own hostname.
All info about the machines is kept in a database, presumably by asset tag number. It would be nuts to try and overload all that into the hostname.
Before my part of $LARGE_COMPANY joined said large company via acquisition, our naming convention was userid-desktop or userid-laptop. That was not a bad system, but maybe not really a good one, either. The current system scales a lot better.
The machine should be reimaged when it changes hands, so resetting the name will add about 5 seconds to the setup process. Not a big deal.
Where I worked before we taged ours based on office and location. If you looked at a sky view of the entire office and were given the computer's name you could find the computer based on the grid location. An example is GM25F. GM (means something to us), 25th collumn, 6th row. That way when we get a trouble report or whatever we didn't have to call anyone to determine where they sat since our maps could tell us just by computer name.
For those who seek perfection there can be no rest on this side of the grave.
you're not sacrificing enough cats
Name them something a user can remember, even if they don't know the origin of the name. I like names from video games. Star Trek names are also fairly popular. Then keep the names in a database. If a user has a problem with the computer then there is a higher likelihood that they would remember the name even if they cant get into the computer. Then just look up the name in the database.
Asset tags....hahahahahahahaha those are great gag stickers!
To ask this question..tsk...tsk...putting all that info into a name...hahahah......it should be obvious to someone with decent IT experience that this is a really dumb question.....I thought slashdot was for geeky IT guys and not guys who are faking it.
cc
If, as I expect, you're working with Windows, there's an obvious naming convention for workstations: Start with Titanic, Yamato, Musashi, Edmund_Fitzgerald, Arizona, Yorktown, Bismark, Monitor and go from there. The theme? Sunken ships. There's an endless supply, and somehow, it seems appropriate for computers that are expected to "go down" several times a day.
Good, inexpensive web hosting
I think I first read it on talk.computers in 1984.
The cake is a pie
Just for reference: RFC 1178
http://www.faqs.org/rfcs/rfc1178.html
While it is not a direct answer to your question, it does give a lot of good why and why not's on this subject. Just as handy now as in the 90s.
We had the fairly sensible approach of doing department abbreviated names, followed by room number, followed by a or b (the rooms did not have more than two, but if they ever did, c and d are waiting), so our graphics department would be like "gfx321a," so when we submitted a problem report to IT, they knew which guy needed to handle it (by department) and exactly where it was.
So, being in a whimsical mood, our supervisor dictated that we were renaming all the computers after superheroes. While this made no sense, the important information was kept in a database anyway, so no big deal, right? Except now IT guys had to look up computer names before knowing where to look for them.
The order of choosing names came in order of longevity (people there the longest chose first). Then, while this was happening, our overlord company demanded we put a two letter company code in front of all of our computer names, ruining our already stupid computer names. Could you imagine having "greenlantern" at Slashdot, and then having to ruin it by putting "sd" in front of it? "sdgreenlantern." [eyes rolling] Wow! That's so cool! You guys are so hip and cool! ugh..
Stupid, sexy Flanders.
I prefer a naming system that when you're in the 'know' it all makes sense, but to an outsider its a bit mysterious. Here.. .I'll give a favorite of mine:
billy, cher, bell, bill, bird, boat, brain, crow, eye, fight, head, horse, light, master, match, mate, pit, roach, shot, shut, shy, stone, sure, tail, up, weed, flood, game, hay, jed, log, pea, pet, and weather :)
In the tightest companies I have worked for, they name workstations and servers with meaningless random generated alphanumeric sequences.
I guess they consider it more secure, making it harder to figure out the network topology. Also, since the names are meaningless, there is never a need to rename the machine really, unless they would want to confuse even more want to be hackers.
Everything I write is lies, read between the lines.
We use system serial number to generate the hostname during the sysprep phase. This is a great scheme imo because naming is based on something burned into the bios, making asset management much easier and it discourages the use of workstations as ersatz servers.
We've cycled about 250,000 workstations through this system since 1999, and haven't had a name collision yet with HP, Dell or IBM
Conformity is the jailer of freedom and enemy of growth. -JFK
One requirement is a good supply of names so you don't run out.
Futurama characters: There are plenty of them, and you can be clever.
City names: you'll never run out. You can assign geographic regions to subnets.
Make them clever and mnemonic in some way! That way you won't confuse w1230fc2 with w1320fc2
In our place, they do both, name the individual office workstations by a more personal Location + CorpUserName combination (which is easy to remember if you know a person's name) and they name the servers with the more typical cryptic Location + Type + ID, etc.
That way, we could easily remote to our PC's or get a shared file from another's PC easily. We usually use only a few servers in our project so the names are easy to remember. It's a win win since they reimage PC's upon turnover and we only need to memorize the server names.
Follow the Naval Aviation standard and name them after beer
Whether you divide by floors, or regions or whatever, just pick a franchise for each group. That way you can tell where a machine is by whether it has a Simpsons name, or a Star Wars one, or a Marvel or DC hero name or something. Make the servers villains, name the printers after those occasional heroes that show up for an episode and are never heard of again.
Admit it. You post strawman arguments as AC so you get modded Insightful for refuting them, rather than Troll
We categorize first by state, next by office in that state, then by what type of machine, and then sequential numbering of those types of machines. So for the 2nd office in Texas we could have the following: TX02W001 - Workstation TX02V001 - Virtual Machine TX02FS01 - File Server TX02P001 - Network printer And then at the corporate level, we have the company name and shared resource type: ABCXS001 - ABC Company, eXchange server ABCDC001 - Domain Controller etc..
First character identifies the type of computer. L for laptop, W for desktop/workstation (D is used for domain controllers) Second and third characters are for the unit code. For example, DE could be your Detroit office The last two characters indicate the year the PC was built (09 for 2009) The remaining 9 characters in the middle (our names are limited to 14 characters) are first initial and last name.
This tells you who the computer belongs to and how old it is. Nice for determining replacement cycles.
My unit within my company does things a little ass backwards from the rest of the corporation. They use user ID's instead of first initial and last name, and then add the last three characters of the PC's S/N to the end instead of a two-digit date code. Still get a nice result to quickly identify the owners of PC's when they start causing issues.
Place I worked at previously had an even much simpler method: the hostname is the cubicle number followed by the image build number.
It made a lot of physical services such as repairs and upgrades much faster and really, there is just too much information about a user and machine to even consider using the hostname to store it all.
You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
We just use our company name and a sequential number. (Some machines like Automated Test don't have service tags numbers)
eg MYBIZ001 then we put a big sticky label from a label printer somwhere easy to see on the PC/Laptop with its number.
All the crap that came with the PC goes into a manila folder in a filing cabinet with the name on the folder.
Any other data about the machine is stored in a spreadsheet.
46137
My scheme has always been Room#_Computer#
The rest of that information can be stored in a database.
*DrugCheese rants*
I assume you're talking about the DNS names, so why not embed all that useful information in the IN TXT field, where the DNS RFC states that such information should be, and provides a huge dataspace in which to put it?
there are enough characters and realms to keep your dns/ldap/kerberos system growing for years to come!!!
Your co-workers can definitely help out with this one. You know those tricks that people use to figure their stage/porn name, take your middle name and the street you grew up on, combine it, and voila, instant new name! Go to each workstation and do that with the person that is sitting there, instant workstation name!
Always consult a standard. For instance, ISO 10992a states that a machine name should be constructed by combining the name, age, sex, and favorite sexual position of each user on the computer, combining into a Unicode string, and taking the md5 checksum of said Unicode string. The resulting hex string shall be used as the workstation name. In the event of a collision, the sexual position of every user shall be replaced by their next favorite position until the collision is resolved.
Our naming convention as is follows. X_XXX_12345
The first digit is the type of computer (L for laptop, D for desktop, and T for tablet). The next set of digits is the three letter abbreviation for the bureau, and the 5 numbers at the end are for the last 5 digits of the serial number of the machine. We have it logged when each machine according to serial number goes out of warranty, changes hands (if we're told, this gets out of hand every now and then), and also helps us identify if a software package needs to go to a machine via SCCM. Works ok but if I had time I'm sure a better one could be thought up.
Starting with "Thomas" a cat I had when I was three and only can remember thru my mothers stories.
There goes THAT tried and true naming convention!
I am very small, utmostly microscopic.
At the company I work for we use the following scheme:
All computers are prefixed with COM to make them easy to find in lists
People who are "permanent workers"
most of these have been around for years and will be around for years to come
computer named with persons name
Computers for relatively temporary people (like customer service are named by location
example: COMCUSTSERV1, COMDESIGN21, etc.
In every company I have worked for, there has always been a naming convention. Now for my current employer:
CA02CS09M
CA = Abbreviation for city
02 = I have no clue, I'm just a programmer Jim!
CS = Computer Services
09 = Nth number in the department (I'm going to assume per purchase/issue).
M = Mobile/Laptop
Let the users name their own workstations. And keep a database/spreadsheet of MAC address, office location, user, and model/serial/warranty information. If you want to get fancy you can update the MAC to user association when activeDirectory logins occur. Then with any IP you can look up the MAC, and know everything you need to know about the machine. You can also quickly determine which machines are on your network that are not part of your IT database. If you have a particularly good managed switch, you can SNMP the port number to find what jack it is plugged into. Obviously location is not very useful for wireless, but you can still setup activeDirectory associations and track machines on your network that way, it's quite useful.
“Common sense is not so common.” — Voltaire
How about famous Mathematicians/Computer Scientists (Church, Turing, Hoare, etc...). Or, famous robots (Skynet, HAL, Keanu Reeves, etc...).
It's fun, nerdy, and educational!
My old university/job used a three letter department code, and then the last six digits of the asset tag. You'd get systems like ITS-26301 and MTH-31415.
This is pretty solid, especially because:
Your mileage may vary.
One word: TinyURL.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
If are Windows workstations, i would name them to designate the net they will be into. R2D2, C3PO, Bender, DARYL, Twiki, to name a few, you have a plenty to choose from. Please, not Data because is (a) running Android.
your-sister
your-brother
your-other-brother
your-mother
your-father
your-aunt
your-uncle
your-cousin
your-niece
your-nephew
These make great pet names, and probably make some very odd conversations in the server room.
...is to use the md5 sum (expressed in sexagesimal, of course) of the MAC address of the secondary NIC, prepended with an "n" for legibility. Simple, logical, and useful.
"I am Dr. Freud, but you may call me.siggy."
We talked about naming workstations after dead rock stars (after all, there's a never-ending supply), but ultimately we settled on elements in the periodic table. One nice benefit is that each had a well defined 1- or 2-letter abbreviation.
Back when we were a much larger company we used car names.
My first Sun box was model-t, which was an accurate description. My next box, a much sexier Sun machine, was twingo. When I set up a box that was a little different (the first Linux box in our department) I named it after a different car, tatra.
My current development box is monaro - a little crude, but very fast. Our new server is the fastest car our sysadmin and I could think of, veyron. It replaced kenny, which really did die one day.
...laura
It's awesome, and you can learn about global mythologies at the same time!
http://www.godchecker.com/
There's plenty of names to go around.
Laugh while you can, monkey-boy!
Name them all Bruce, just to avoid confusion.
Pick a theme and stick to it. In the past, we've used names of fish and that seemed to work pretty well...swordfish, hammerhead, etc. Obviously, pick something with a sufficiently large set of names for your network. This is effective as non-meaningful id's, but gives the workstations the sort of personality that they deserve.
Avoid using information *about* the workstation in its name. Primary user, location, function may change over time. Renaming is problematic so avoid needing to do it.
We do (WS|LP|MAC)_machineserial where WS is Workstation LP is laptop, and MAC is well MAC machines. Then we use Landesk to gather current user and software configuration.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Admins are idiots ... sucking the spickot of presumed power.
Management is brain dead and seeking a bed for comprimising the company for dollars for F*cks per hour. Like a 14 year old girl, or a horse, they may not have much of a brain, but they know that everybody wants to F*ck them.
I would not have thought that machine names could arouse such a passion.
Why Should I Care What Color the Bikeshed Is?
From http://www.freebsd.org/doc/en_US.ISO8859-1/books/faq/misc.html#BIKESHED-PAINTING:
"The really, really short answer is that you should not. The somewhat longer answer is that just because you are capable of building a bikeshed does not mean you should stop others from building one just because you do not like the color they plan to paint it. This is a metaphor indicating that you need not argue about every little feature just because you know enough to do so. Some people have commented that the amount of noise generated by a change is inversely proportional to the complexity of the change."
minion1, minion2, minion3 ...because that's all they are.
I use Sushi dishes in my lab. nice short names :)
Uni
Sake
Unagi
etc
Can't give you any advice since I'm still struggling with this but can say...
TEMSEXCHANGE01 is BAD CHOICE!
I don't give a damn if your company name starts with TEM and it's the acceptance standard!
There's just some about the following that's wrong...
TEM - company
S - Server
Exchange - Should be obvious to any self respecting geek
But for christ's sakes...
Tem-sex-change-01?
Users, tech support, and admins can easily remember names and lookup details in a spreadsheet or db. At the prior bank, workstations had names like 'piglet'. When looking at DB locks, the first few times I had to look up the user, but subsequently I quickly memorized the frequent workstation names I came across. Where I work now, they use an alphanumeric identifier. I can't even remember my own workstation name. Workstations already have unique id's like a MAC address and service tag - the machine name doesn't need to be hard to memorize like these are.
Now, that's a hell of an idea.
Understanding the scope of the problem is the first step on the path to true panic.
We just name machines by their serial number. It's so easy. want to know a machine name? it's right on the front of the machine! Want to move the machine to another office with different rights? Just drag and drop it into a new OU while you are at it. Any other info goes into the description.
That's a lot of work when someone changes a cubicle.
There are more than enough of these to handle the largest data center. Start with cthulhu, vaal, landru, mcp (think TRON). And don't deny it: you've named production servers after most if not all of these names, haven't you?
The server was named after a main river (Columbia). Tributaries that produced fish (deschutes, snake, johnday, hood) were production machines and dammed rivers (whitesalmon) that had no runs were isolated testbeds.
This was in 1995-2000, So YMMV.
I prefer to name my machines after Pokémon.
Seriously, though, I go for something descriptive. My machine here at home is named sheeettin-kubuntu (because it's mine, and it runs kubuntu, and I don't use the hostname that often).
At school, they use a room-ID system, e.g. the first in room 210 is 210-01. (The server I administered in my computer tech class was named "poopserver", both because I needed a name during setup, and the server ended up serving crap like Counter-Strike, Gameboy Advance ROMs, and Dragonball Z episodes.)
Like others, we use asset tags (or last 6 of serial number), but add a prefix to ID the operating system for the Help Desk and to help with SMS reports. For example, XP for Windows XP and 6W7 for 64 bit Windows 7. For VMs, we add a V1, V2 suffix. We have an inventory database that correlates tags with PO, warranty, vendor, etc. It's working fine for about 4k machines.
We use the following Location-Department-Asset Tag For location and department fields we use a two to three letter abbreviation. So a Customer Service desktop with the asset tag of LKI13 would be: DEN-CS-LKI13 For laptops we substitute LT for the department so: LA-LT-LK9F7
Ape, Bat, Cat, Cow,Dog, Doe, Elk, Eel, Fox, Gnu
Hog, Hen,
Imp (ok it's not real...)
Kit (Baby fox?)
Mus (Latin for mouse)
Ram
With one or two exceptions the Computer Science Club of the University of Waterloo uses these.
Each position is meaningful, making the entire name even more meaningful.
1st char is workstation manufacturer initial.
I is for IBM
D is for Dell
A is for Acer
H is for Hewlett
L is for Lenovo
A is for Apple
2nd digit is the floor number. 3rd and 4th digit is the room number. 5th position is the processor type I(ntel) or A(md). 6th position is the location of the user manual. L(ibrary), D(esk), S(torage) or C(ircular file). 7th position is W(orks), B(roken), P(arts). 8th position is P(orn) or C(hurchlady). 9th position is A(nniston) or J(olie). 10th position is 3, 4, 5, 6, 7, or C(an't possibly be that large you liar.)
My own computer is conveniently named L202ICWPJC.
Fascism trolls keeping me up every night. When I starts a preachin', he HITS ME WITH HIS REICH!
WG == work group, ### = work group number
WS == workstation #### = work station number
Keep everything else in a simple-minded database. That's about the only practical solution for a large organization with hundreds to thousands of machines.
OTOH -- years ago when I was at startups, we had name "themes" per work team. The "Jay Ward cartoon characters" net was great -- watching the VP's squirm every time they had to say "Biggy Rat" with a straight face was exquisite.
Another net was the "beer" net. I specifically asked for 'leinenkugel' because if someone was looking to steal some mips for a simulation, they were far more likely to rlogin to 'bud' than a machine they (being an engineer) couldn't remember how to spell.
I name my home machines after baked goods. Cupcake, Donut, etc.
Every engineering cluster had a theme. That meant that you knew what lab the machine was in but it still kept the names interesting. It also made it easy to remember that the dolts who killed remote jobs always used the NBA team machines because their prof told them to use that lab and how to kill processes.
The best theme? Rain, Snow, Hail, Leaf, Meteor, Skylab, etc. "Things that fall from the sky."
Ram rod
Stonewall
Cathode ray
White Swallow
Manhole
Mineshaft
All of the "Johns" from Buckaroo Banzai. Worfin, Ya Ya, Small Berries, etc. That is all.
#include "humorous_pop_culture_reference.h"
At my workplace, computers are simply named by type (PC for desktop, LT for laptop, etc), followed by a batch number to indicate the model and purchase date, a dash, then a unique four-digit number. It will turn out something like PC02-0781 or LT06-0030. Staff swap computers regularly, so there's no point in tracking them through the computer name. All other information is stored elsewhere.
Marklar if you are a South Park fan, Zathras if you are a Babylon 5 fan. Too bad neither of these scale too well.
Any insufficiently advanced magic is indistinguishable from technology.
I name my computers after Greeks real and mythical. Most of them go by a single name with a simple, phonetic spelling.
The desktop connected to the stereo used for playing music and video is Apollo
My grandmother's old, slow laptop is Thales
My more recent desktop is Pythagoras
My Macbook is Euclid
My iPod is Orpheus
My VMs are all named Plato
There are several famous real and mythical Greeks, so I don't risk running out of appropriate host names.
I'll be your candy shop of infinite deliciousity if you'll be my discotheque of endless rump-shaking.
Open a dictionary and pull every word that has 8 or less characters...there's your list of computer names. Any identifying information in the computer name should be limited to a two or three letter suffix that represents the building, site or city (for those with multiple sites within the organisation) or some other identifying mark....but keep it 3 characters max. A good pre-made one is airport codes....three letters...tells you which city the computer is in. For laptops the suffix represents the 'home base'.
If your organisation is large enough, everything else should be in a database.
words are easy to remember, easy to spell and easy for the users as well. Get a dymo label printer and print a 3-4 line label with hostname, ip address/MAC address, asset tag (where I work we also include the lease end date). Stick it on the side or on the lid (or somewhere discrete but accessible) Keep it simple, keep it short....for all intents and purposes the host name is just a descriptor. Everything else should be in a database.
I like something allegorical. Our company is Endeavour Partners; we use the names of Captain Cook's ships. My previous company was Mercator Partners; we used contemporaneous names of countries for servers, of major cities for desktop machines, and of ports for portable machines. I'm the CTO of a startup involved with horses: servers are racetracks; workstations are famous racehorses.
Given the amount of discussion, this is obviously not a silly question. I worked at a company in Palo Alto for a while, and one of the things that appealed to me was that they let users name their own workstations. You got a completely random mishmash. They didn't reflect the machine's purpose (more secure, if that worries you); they're easier to remember (betty, veronica, larch, elm, etc are way easier than random alpha strings); you don't have to change them if someone moves; and when someone quits you reimage anyway, so let the new owner choose a new name, or stick with the old name if you prefer. It's also more empowering than some faceless (and finite) naming convention imposed by the trolls.
Start with 7 and keep going till you hit a billion. Then leave anther six and go till two billion. If you don't like 7 start with a number you like.
A good unique identification scheme should not overload names with identifiers/location/sexualOrientation, etc. Just provide a unique identifying number and keep all data in your database. Sync the database to your pda so that you can pull information up even on the move about the machine when you get a user request.
O this learning! What a thing it is - William Shakespeare
Jukebox Hero is my media server. Hot Blooded is my laptop that tends to overhead. Head Games is my gaming box, Double Vision is a VM I use to test websites. Urgent is an IMAP server with the push extension installed.
There are two parts to our workstation names at our small company. The first part is the asset tag number, usually 3 digits. The next part is the intials of whoever that computer is assigned to. IE whose cube it is in. For example, 437-JND for asset #437:John Nathan Doe
Just a character with a serial number.
All other information can either be automatically gathered, and thus should be, or belongs in an administrative CMDB. And b.t.w. the same holds for servers.
Really: just a number. If you don't do that you shoot yourself in the foot with having to maintain information in a place where it is inconvenient to maintain, AND you risk administering the same information twice, which is a burden, and a source for errors.
At our company we give every computer an internal serial number in the form of a letter and four decimal numbers which becomes the hostname. A computer gets a sticker with this serial.
We also have a custom database where we can look up all information about a particular computer given a serial number or any other information (username,...).
Number of computers: 300+
Works great and can't think of anything that can currently beat this system. It's future proof since we never need to rename a computer no matter what.
My 2 cents.....
If your company has any kind of fixed asset tracking then there's an asset number label on the machine. So just use that as the machine name and keep track of things in a database. Easy; the end user can look at the asset tag when the help desk wants to know which machine and you can spot them on the network easily.
You're going to have enough trouble keeping track of what machine was assigned to who without changing machine names at the drop of a hat. And this will reduce the number of places where information is mis-keyed or just blown off as too much trouble to one - the asset to user database.
If you encode user or location information into the machine name you'll have two things to update each time a machine is moved. And you'll discover that your end users swap machines and take "their" computer along when they're transferred to another location. Not to mention the department managers that see that the new machine one of their drones got is much better than theirs and swaps them. None of these people are going to call in with updated information so you're going to have to track that in a different way.
If you want to track locations, then how having multiple subnets on your network - each location is in a different subnet. That solves the mysterious moving machines problem because the machine will have an IP address that shows what location it's connected to the network in.
Just to keep it clear.
This issue is a bit more complicated than you think.
We name the computers as follows: -Asset Tag number We have an asset tracking system where we keep track of all our equipment. In that system we keep track of when it was purchased, when the warranty will expire, etc.... We name each image with a name that says some thing about what image it is such as XP2 or XP3 or our latest one we called it DM5 for the version of our document management system. So if we have an HP dc7700 computer with the DM5 image and an asset tag of 12345 the computer name would be DC7700DM5-12345 So we can easily look at it in AD and see what model and computer image is on that computer, and if we want to know more about the hardware we look up 12345 in the asset tracker.
Curious about Storage and Virtualization? Check out
Name them according to the botnets that they are members of, and use whatever unique identifier the botnet assigns. This may involve a wait of a day or two before they are enlisted in a botnet, but the bonus is that some machines will acquire several names before long. This scheme really only works for Windows boxes, so you'll have to extend the naming policy to ban Linux and suchlike from your workplace (thus ensuring Microsoft's approval for the naming policy).
Those who can make you believe absurdities can make you commit atrocities. - Voltaire
Name all your machines "Eric".
My site
Looking at all the replies and the discussions, the OP didn't ask such a bad question.
At our company most of the machines are named after famous movie stars, except for the top end machines are named after japanese porn stars. You may be saying to yourself 'what an juvenile, amature setup'. Surprisingly enough, we are a VERY big company (a couple years ago we ranked in top500.org 's most powerful data center list). It amazes me that this fact has escaped management's notice for so long...
Most elegant system I've seen simply uses incremented numbers. They started at 1....currently they're into the several hundreds I guess. They maintain a separate database that indicates what Machine #23 is, who uses it, where it's physically located, what OS/major software revs are on it and ties any service calls to it as well so they can see that users of Machine #23 frequently complain of trouble printing to the network printers, for instance.
Clean, easy, elegant, when the machine moves to a different desk (as they sometimes do) or gets upgraded or gets a new user or a new purpose they don't have to do anything to the machine name. They just update the database and go on about their day.
-B-
All of my workstations are only named after a peyote induced vision quest. After that the machine names themselves.
Up to a couple hundred people, I like UsernameAssetTag. I also like to reimage a computer when it oges to a new person.
It is very handy to know the primary user of a particular computer. Having the asset tag in the name is handy for tracking computers.
I use a code system for our 300 workstations, eg PC021OAP = computer#21/OFFICE/ACCOUNTS/NETWORK PRINTER. This identifies its in the office building, accounts section and has a shared printer attached, and is unit unmber 021. While this may be limited to your application, it provides the administrators enough info to control the unit regardless of who is using it.
How comfortable are you, if strangers can reverse engineer the structure of your network? What if they can identify the laptop by listening to broadcast queries at a public Wi-Fi point? The more information you place in a name, the more you expose.
The higher your paranoia level, the more you want to stick to random names.
PEBKAC_0001...
American , I believe, name ships in the same group all starting with the same first letters - George, goober, and Gork. The british name thiers after a common theme - Hatchet, Axe,Chopper - I dunno.
So conceptual vs similar sounding - some work better than others for various people. Not one solution is perfect for everyone.
..........FULL STOP.
A name needs to be recognizable by humans. Because inevitably someone is going to want to share some files and it's a whole lot easier if you can type in a normal name instead of mistaking RS34598 with RS34589. Granted, the user's name isn't good, because machines change hands all the time (without telling the busy bodies at IT about it). Cube numbers don't work, since a lot of machines are lab machines, or may turn into lab machines.
There really isn't a good way. Would be nice to have two names, a permanent one, assigned early on, probably related to an asset ID, and a nickname based on the user or purpose of the machine. The nickname can be changed anytime the user or department wants to do so. Except that this may be a pain to do on some operating systems.
True, at work it takes me a whole 20-30 seconds to rename a computer when we move it to another room.
I named my machines as groaningly as possible. My desktop is named Courtney (a combination of Core 2 and Tony). My apple laptop is simply "Roving Pear." This may be harder for a large network, but I bet you can manage.
In my homenet, I have:
SAKURA - Nintendo Wii (Sakura is a well-regarded Japanese cheese)
EMMANTALER - Apple Airport Extreme-N (Emmantaler is a decent Swiss variety... lots of air in those holes)
MINI-BABYBEL - Apple Airport Express-N (the two Expresses are a pair, Mini-Bonbel and Mini-Babybel. if I get another one it'll be Mini-Gouda)
MINI-BONBEL - Apple Airport Extreme
GRUYERE - Apple Mac Mini (Gruyere is excellent in mac-and-cheese)
EDAM - Maxtor Central Axis (no particular reason for this one)
STILTON - Dell Dimenson 4300 (no particular reason for this one)
BRIE Dell - Inspiron 600m (no particular reason for this one)
GJETOST - Apple MacBook Pro (Gjetost is a Norwegian brown cheese that is really good with sliced apples)
SAGEDERBY - SageTV HD Theatre (SageTV -> Sage Derby cheese)
DOUBLE-GLOUCESTER - Home-built file server [retired] (it had two drives, thus Double)
Naming is extensible practically forever (how many different cheeses are there?) and while some of the names have whimsical meanings, some don't and it isn't necessary. Names are mapped to machines/devices in a spreadsheet. And it's unique among everyone I know. Lots of people use Star Trek or Tolkien or WoW, but I don't know anyone else who uses cheese.
Vista:XPSP2::ME:98SE
RFC1178 still contains some useful advice.
The manufacturer or supplier has already provided a useful, unique ID: the Service Tag. Just keep track of them yourself for accounting and support purposes, and the Service Tag gives the additional benefit of providing the self-supported end-user with direct access to correct support on the suppliers' web site.
Who?
We named the machines after colours.
There are already tables to translate colour names to RBG and vice versa.
RBG can be easily derived from the address (10.RRR.GGG.BBB).
Grouping them by rooms, gives a red room (crimson, ruby, salmon,...), orange room (orange, tangerine, carrot,...), ...
Red room is students lab. Administration is on ultraviolet.
The general rule is "keep it simple". Of course, that depends on a lot of things:
a) how many users you have
b) what your purchase cycles are like
c) what kind of user/system pairing you've got (ie 1 system for 5 users, or 5 systems for 1 user)
d) how many 'departments' or work-units are we involving?
e) possibly most importantly, how much of a pain in the ass are your users? IE, are they going to bitch and moan if they don't have "johns-computer" at login (like they're used to due to prior poor management)?
Personally, I try to make workstation names as short and memorable as possible for management purposes while still retaining a degree of clarity telling me where they are on the network. Again, it all depends on the specific environment to a large degree, and seems to get more the most complex and frustrating around 20-100 machines (due to organizational momentum and poor planning).
I like to put each 'working group' within its own domain, and depending on the size of the work group, name the computers after a person's position and role. For instance, manager.hr, clerical1.accounting, and so on. It eases application and GPO deployment through AD.
In places where I've come upon many small purchases (eg. 200ish machines, but all purchased 10 at a time) I'll put the batch number in the name: eg. b11manager.hr, with the b## being a number that increases with each new batch. This helps make the environment somewhat self-documenting, and is useful for when nobody updates documentation.
~/ssh slashdot.org ssh: connect to host slashdot.org port 22: too many beers
Just make the name a unique identifier and have done. You can name them after plants, animals, politicians, authors, performers, composers, mathematicians, scientists, cities, or consumer products. It doesn't matter how you generate the names, as long as each computer has a unique name assigned when the computer first enters the organization and retained until it leaves the organization or is scrapped for parts.
Resist the temptation to put any information in the name that might possibly change before the computer is permanently retired. It's tempting to put user or location information in the computer name, but this is a bad idea. Put that stuff in your inventory database (or, if you're a small operation, just make a list in a text file) so it can be updated without renaming the computer.
Cut that out, or I will ship you to Norilsk in a box.
Naming for Workstations should be the same way as with servers. Meaning the name should be distinct and not be associated to the function of the workstation or the OS installed. If an Error message pops up I should imediately be able to recognize the computers name. If the box is called 'Linux' and I read 'Linux error' I'm likely to be confused about what is meant.
I've found it best to find a broad naming scheme like Astronomy and go by and name all servers and workstations based on that scheme. 'saturn', 'uranus', 'antares', etc. If you must you can go by and name the servers after planets and the workstations after moons or something like that. Allthough that in itself could allready pose a security risk if some hacker sees the pattern.
We suffer more in our imagination than in reality. - Seneca
I've found that if it isn't automatic it gets forgotten.
Three years later you'll have WRKSTN_ROOM423 in room 132 and the admin or user that moved it will have either forgot completely or moved on.
Workstations should not need to be accessed over the network so they should not need a friendly name.
There is no reason why the tag number which is clearly printed on the machine should not be used.
Do you also give names to your phones?
Your photocopiers?
Your water coolers?
Computers are just frigging machines, tools, just like a hammer or screwdriver (do you give names to those?) only immensely more complex.
As soon as you have more than 10 machines under your responsibility, silly names become actually unproductive (you waste more time naming a new machine than actually needed).
IANAL but write like a drunk one.
You have to rename it.
Which is silly.
As with people, machines should have a unique name, all the rest of the information about the machine should be in a database of some kind (a list in a text file would do).
Then when you move the machine, assuming that your DHCP, DNS and WIntel servers are up to scratch, yo have to do precious little but relocate the machine (and update your database).
With your naming scheme you have to rename the machine in addition to updating any database you may have.
IANAL but write like a drunk one.
For me a lot of machines means 500+...
IANAL but write like a drunk one.
A similar issue was brought up in one of the stories linked in 'previously' - conference rooms named after peaks higher than 14,000 (apparently, the "fourteeners"). Well...I have one example that's worse in implementation - PLANETS. At an office site in Malaysia, one of the floors has meeting rooms named after planets...in our solar system...
.
So when someone told me, we had a meeting in 2 minutes at Neptune, which I hadn't come across, I grabbed my lappy, power cable, charger, writing pad, lan cable and rushed off in the opposite direction as Mercury, assuming they were NOT next to each other. Right? Wrong!! Some retard decided that shouldn't be the sequence of rooms, coz we'd apparently not remember the goddamn order: mercury, venus, shithole, mars, jupiter, saturn, uranus, neptune, pluto (even to an approximation if required).
.
I turn around and run with all this crap in my hand, to the other side of the floor next to Mercury! Neptune was next to Mercury! Pluto they decided should be next to Jupiter. Later on, I realised it was related to SIZE. Mercury & Venus were both 4 seaters, Neptune was an 8 seater, Pluto was 2 and Jupiter was 10+. Yeah, real clever! google reference1 & reference2.
.
So we're not expected to know the order but the designer wanted to educate us/others on size? How very useful. Mercury & Venus are not the same size and there is no 'Mars' or 'Saturn' room. I can understand skipping this one and Uranus.
.
I'm sure the designer has it scratched down somewhere, on which date the planets would be aligned to rooms' order when looking at the solar ecliptic. Fucktard.
Workstations should be named after the person who primarily uses the system. Unlike servers, specifics like this are okay, because you generally do not have a lot of things set up to work for the machine as much as for the user of that machine- and you usually don't mind if those break when staff changes. Workstations are okay to change the name of. If you find yourself in a situation where you can't easily change the name of a workstation without breaking something, you're doing something seriously wrong.
Small and catchy names, like naming things after cities or animals or whatever, as others have suggested, is a horrible horrible stupid idea. Catchy names are only good for resources which are shared in a network environment- which workstations should not be.
-- 'The' Lord and Master Bitman On High, Master Of All
As I said below, my company does the same.
e.g.
ebtgp3 or iydbdg
Read them out, over the phone, and enjoy the sick humour.
This post contains benzene, nitrosamines, formaldehyde and hydrogen cyanide.
Hydrogen, Helium, Lithium etc...
You could even associate the electron count with the final digit of the IP address or something....
Yes, I did that in all caps, because it doesn't seem to sink in for most helpdesker/phone-droids for wanna-be sysadmins that plauge IT there.
Hostnames are not documentation.
Hostnames are not documentation.
Hostnames are not documentation.
I hate places that try to name servers/workstations after some stupid coding methodology that ends up with "CLSmr1f18" for a hostname.
Also awesome(as in stupid) are the people who don't know about domain names and put the domain name in the host name. So, we have "mycompany-server00001.mycompany.com" Their next server is named "server2_mycompany.mycompany.com", just so they can score for inconsistency madness.
Other idiots insist on zero-padding all the server names (some with just a single zero, some with a double-zero, some with none). So, we have fifty servers that all end in 01. Like thing-web01, thing-app01, thing-miami01, otherthing-app01, zippy-app01, dumb-app01, fark-web01, etc etc etc. We have absolutely no server out there named anything-02 or above. (for added fun, use underscores in some hostnames instead of a hyphen). Zero padding is for computers, not people.
RFC1178 should be retired reading for all IT staff on their first day at work. Though, I would definitely like to see RFC1178 appended with the simple statement, as above;
Hostnames are not documentation.
Seems like an increasing number of places are naming workstations after the employees whose desk they are on. This not only helps IT identify where they are, but also identify what box your coworker is talking about when he says "I've got it up and running on my machine."
Others simply give them a brief categorization a serial number, like LAP345 or WKS456.
One place I worked years ago that made digital reference books named their workstations after sequential words out of one of its dictionary products. This was a little strange as pretty much every workstation name started with A.
Terrorists can attack freedom, but only Congress can destroy it.
It is a common mistake, but do not attempt to insert descriptions into identifiers. You wouldn't name your child "Dribble-gums-nursery-2" and expect then to be still comfortable about it when they reach their teens. But call then something meaningless like "Kevin" and there's no problem. Computers are no different.
If you create an identifier that attempts to describe the computer, rather than just give it a unique name, you can be sure that by the time it comes to decommissioning it the identifier will be misleading. Things will have changed. It will have a different location, a different OS, a different owner, or a different spec.
The naming convention I use might not scale to well, but it is perfect for the network I manage.
On this network there are three things that rarely change over time, the owner/user of the computer, the operating system, and the purpose of the computer. The names I use are therefore of the form XYZ-NAME, where X, Y and Z are single characters that shows who owns/uses the computer, what operating system it runs, and what the purpose of the computer is (in that order). NAME is the actual name of the computer, this is usually an animal name that I find appropriate.
So Bob's server running Linux would be called BLS-CHEETAH (BobLinuxServer-CHEETAH), and John's workstation running Windows would be called JWW-DONKEY (JohnWindowsWorkstation-DONKEY).
Plenty of name generators on the web, such as http://www.seventhsanctum.com/. I quite like the dwarf names such as Bloodbreaker, Demonbreaker, Doomsmelter, Foesmiter, Greatmail, Honorpick, Irondig, Ironsmasher, Lightpacer, Stonebullion. One serious advantage of generated names is that they are pronouncable, making help desk support easier. Unlike some alphanumeric codes - I still remember the confusions when IBM had two RS/6000 family members, the 380 and the 3AT.
Andrew Yeomans
At our University the Computers are named after characters in 80's TV-Series
The MAFIAA is a bunch of mindless jerks who will be the first up against the wall when the revolution comes
Why not name them all Marklar. Keep it up for other stuff and soon you'll be able to reference to all things, people and places as Marklar. You'll never have to worry about names again!
(for lab computers) ... vertex60. pixel01, synapse01, glyph01.
Pick something computing/science/maths-sounding. Name all computers of the same type with that, plus a number: vertex01, vertex02,
It's not as boring as "asset1241", but it's a *lot* easier to find numbered PCs in the lab. It's also easier for anyone wanting to use a machine remotely. Finding your usual glyph12 is running slow? Well, you know at least 11 other machine names.
Staff/research students could name their own PCs, presumably because it's a lot easier to find one PC out of just three in an office.
Servers were named after birds, supercomputers after (IIRC) greek gods, and the authentication servers after nuclear accidents ("there's a problem with three-mile-island, so I've changed the DNS to point to tokaimura")
You have 2 requirements - accountability, and ease of remembering. Usually one without the other. Simple.
First, you assign a range of IPs to an arbitrary number, containing a 3 letter location, which never changes, regardless if you rebuild the machine. Eg. for London:
lon-001 A 10.0.0.1
lon-002 A 10.0.0.2
lon-003 A 10.0.0.3
etc.
Then create meaningful names. eg the person who uses the machine, or what it does if it's a server:
dave-pc CNAME lon-001
sarah-pc CNAME lon-002
proxy CNAME lon-003
I wrote my first program at the age of six, and I still can't work out how this website works.
First genuine LOL of the day. Thank you !
Squirrel!
That's a lot of work when someone changes a cubicle.
Is it? How much work? Sounds like a nice plan that should save more effort than it causes. This might depend on how often people change cubicle, but how much work is it really to change a machine name?
If you plan to bother with official names, use the asset tag and use some other tool to retain all the information on user, location, build, etc...
Be certain that the asset tag is also stored in the BIOS, so you can use tools to query it.
The way you deal with all this depends greatly on whether you have 20 workstations or 100,000. Tracking locations for 100,000 desktops is a difficult thing. Heck, tracking locations for 40,000 servers is a difficult thing.
Whenever a workstation is turned over to a new person, do everyone a favor and re-image it.
Right, keeping a spreadsheet of all relevant information is one thing that you should do, especially if you're in a strict organization that needs to keep track of every piece of equipment. Secondly, I work in a university computer lab. Right now, I got about 3 rooms full of computers which are placed against walls and barriers so I give them the lab name, the room name and a sequential number counting clockwise from the door. Hopefully, it'll make more sense to my boss than the last scheme, counting them by order of arrival.
No one ever said being a Heretic was easy.
Let us meet again in "Less Interesting Times"
We used to have [User_Initial]-[Room#]
And yes, the room numbers changed, frequently.
Our new system is [Department]-[User-Initials]-[Modifier] where the modifier is 01, 02, etc if they have multiple machines. Machines that are publicly accessed or are not assigned to just one person are named [Department]-[Room#]. The room number on public machines changes a whole lot less frequently than user machines. Notes are listed in Active Directory with current room and the user's name since we can't remember whose initials.
If we gave a PC to another user we'd just image the computer again.
No - just give them a new machine when they move - with applying some social skills, they'll see it as an upgrade.
This is every bit as stupid as the "how do you name your servers" question. And for what it's worth, anyone who is embedding that much data (site, warranty end date, machine type, etc.) is wasting too much time and providing too much info to potential hackers. I've always lobbied against embedding any kind of specific info into system names because it can change. Lots of companies like to indicate in the server name if it is physical or virtual, which works great until you P2V a server and it goes from physical to virtual without changing the name (because if you're going to rename a server, you might as well rebuild it). The same goes for desktops/workstations with people's names on them...it's all great until Bob get's fired and you give his new replacement named Martha his old laptop. Sure it should have been rebuilt in the meantime, but did it?
Keep it simple. Name your PC's numerically like "PC0001" or something like that. Laptops can be named "L0001". Name your servers "SERVER001". Everything else that you could possibly want in your name is available via WMI (or some version of WBEM/CIM for you non-Windows people). Even if you're not using commercial management tools, just add a line to your logon scripts that polls the relevant data and writes it to a repository. Or you could run a script on a weekly basis that crawls the network collecting said data. Or better yet, both.
We name our by Location, department, and last 5 of mac address. Works pretty well for us. Depending on the size of your company, I would use bldg, room, and last 5 of mac. XXXXYYYYZZZZZ. Just my anonymous two cents.
because if you're going to rename a server, you might as well rebuild it
What, "hostname $new_name" is too hard to type? I mean, you don't hardcode the machine name in application config files and rc scripts, do you?
Do you?
Why name a machine that doesn't come when you call?
I usually use a combination of the site name that the computer sits at, the department it is being used in, and the job title of the person using it. Oh, and each of these things are abbreviated. So a workstation name that appears like CO-AC-PC001 would be Central Office - Accounting - Payroll Clerk 001. Seems to work after you get used to what your "codes" stand for. I find tying the last part of the name to a job title instead of a user's name makes it easier when people come and go in positions.
Our standard is using 10 characters name where first three characters is for location (i.e. NYC) followed by three next characters for desktop or portable (i.e. dtw for desktop workstation and ltw and laptop workstation) and finally the last 4 characters are sequential numbers. Same convention works for our servers (i.e. nycsvr0039)
I work in a school with over 5000 computers... we use
school initial-room#-computer#-model code
IE
HA-202-01-DD09(dell desktop year 09)
that way in our management we can see what kind of computer it is, and the location when we need to remote to them.
syrinx, lamneth, narpet, cygnus, snowdog, bytor, maelstrom, panacea, yyz...
It depends if you are administrating a small company where you make the rules or a large company with directives. For a small company with only one site, naming your machine is worthless. But, for a large company with thousands of workstations and several sites (that have portable workstations moving between them) it helps to have a concise naming system. The best naming system is to have the first 2 letters be relevant to the site's name, then a simple 3 or 4 number account tracked by your supply depot, and the serial number or MAC address (if accounted for by supply) of the machine on the tail. This way you know (through a DNS/WINS lookup and a fancy network management tool like Cisco works) if a person from a different site connected to a port in your site or if a different account with your site decided to move to another part of your site. This is helpful if you need to track down a machine that is being bad even if it is off the network for a while. Good naming conventions are extremely important for good network management.
In small shops, it doesn't really matter what convention you choose as you would likely know each workstation anyway, so pick whatever you like (e.g. mythology, human names, Pokemon, etc).
In medium companies, it might be a good idea to assign unique ID to each machine, and put user/location/hardware information in a spreadsheet or a database. Some minimal information maybe embedded in the hostname - such as whether it is a desktop or a laptop (e.g. DSK1234, LPT5124). Each move or ownership change should be ideally done by IT, who should then update the spreadsheet.
In large companies, there will likely be several IT departments (one for each business unit or branch), so manual update of each change will be a hassle. I think the best solution is to automatically put unique hardware-specific information in the hostname during initial imaging. MAC address seems to be the best choice. Once imaging is done and the machine is deployed, inventory tracking agent should report hardware and software information to a central database, which would deduce machine's location based on its IP address.
Don't bother embedding user or location info in the hostname, as it will cause more harm in the long run when machines start changing owners or locations.
Disclaimer: I worked in companies of all three types, and this naming convention worked well for us.
At my current job, since my office is so small, we use the actual user's user number, an OS code, and a system number (ie if they've had a replacement or if it's a temp machine.) Something like E0000098-XPA. (Vista is VB for Biz, right now, since we only have one Vista laptop in use.)
I had suggested using cube numbers - since it's easier to equate the cube number with the person, due to a lot of "self approved migrations" - but was shot down by Net Ops 300 miles away.
At home, I use ship names, planets, etc from Stargate.
In a corporate environment the asset tag is a good start but with multiple sites and devices you can expand on it with: Site ID-Device-Asset tag The device is usually a workstation (w), black and white laser printer (lb), colour laser printer (lc), etc... This way the support guys at least have some idea where it is and what it is when they get a call. The name is unlikely to change as moving kit from site to site doesn't happen often.
Computer names you!
I like the idea of including the OS image build version... including the machine base model might also be of some value. Then again, some of this is just as well in the machine desacription.
We're a complicated mess over here, multiple major business units, each with several minor units, and each with their own budgeting and asset owership, so on some level we're including who OWNS the machine in the name, and it's base OS and an ID, it might be hard to add build number to the name (but in the description would work).
There is no contest in life for which the unprepared have the advantage.
I'm guessing the computer stays with the cubicle.
Choose a theme (a book or film series, a part of history, whatever), then name all the workstations that are close (work together or in the same room or whatever) after this theme.
We have a network of about 30 machines and our naming convention follows as SVR-###, USR-###, and INS-InstrumentType. (Server, User workstation, and Instrument) Machine information is put into the description so that it only needs to be changed on the server.
I've started using a combination of the operating system name, the computer brand and model number, and a dash/number on the end when there's multiple systems of the same type.
Wanna fight ? Bend over, stick your head up your ass, and fight for air.
In the current ecomonic climate: someone changes cubicle == redundancy. No work required.
We use a name that tells us just enough about where the computer is and what dept it belongs to.
HR01W100 or HR01L800 (Human Resources for ex.)
using the first two as a dept code the second two digits are the building location and a single letter for machine type (L=Laptop, P=Printer, W=Workstation) then a three digit incrementing number. this lets us kow where the computer is what dept it belongs to and what type of equipment it is. It also means that you don't have to change the name per user or room move, only if it moves between buildings or departments. We use this convention to name workstations, laptops, printers, servers, terminals, and even handheld devices. It's very scalable allowing any dept to have 999 of any one type of device in a single location.
Where I currently work, our systems are named with a combination of 1-3 letter codes. Included in the code are building and room, machine type (server, workstation, printer, etc.), and network (we have multiple LAN/WANs that systems can be a part of). Works pretty well to help someone physically identify a system, which is really all you need anyway.
Worst scheme I've seen: facility name, subnet, node. Facility never changed (it was only used there) and subnet/node are already part of the IP, obviously. So having a much of machines named things like "ETC_37_123" and "ETC_37_124" really didn't do a bit of good.
One place I worked, we used a different scheme for each subnet. One net was movie titles. Points to anyone who can tell me why naming a machine "2001" in a UNIX environment is a PITA.
This depends on your environment.
If you have mulitple sites, something like this:
AAA-BBB-NNNN-C
A=Location
B=Optional letters.(probably not needed)
N=Numbers, starting at 1. You can get a feel for the age of the machine at a site based on the #.
C=Machine type, W= workstation, L=Laptop, S=Server, etc.
Typically this would leave you with:
CHI-0021-L
CHI-0099-W
TEX-0198-W
TEX-0199-S
Odds are the 0021 machine should be replaced at some point.
For a single site, if you have the policy of machines being reimaged when being reassigned, you can use names. I am not a huge fan of names on machines.
JSmith-0019-W
MSmith-0020-W
MSmith-0078-L
There are many options, but really the key here is to simplify. What do you need. What do you want. What gives you the best value add. All these are factors only you would know. To answer your question. "Do the management tools in use make a difference in how workstations are named?". These tools can make a huge difference. Generally an inventory management tool will allow you to associate all this important data to the machine outside of the machine name, to the point the name becomes irrelevant. If you can sort a list based on warranty date, location, user, memory installed, CPU speed, it doesn't really matter what the name is.
the only woman who truely broke my heart. Yes I'm looking at you Agent Scully.
My other sig is a knife wound.
We name computers by asset tag + a letter showing if it is a desktop or a laptop.
Last place I worked, we used the user portion of the email a dash then the desktop inventory id. ie, bsmith-E23952
You knew who was using it without looking it up, and you could look everything else up about it with the inventory id. (cpu, memory, location, warranty, etc)
Staff/research students could name their own PCs, presumably because it's a lot easier to find one PC out of just three in an office.
I find that this leads to people anthropomorphizing their machines, and trying to later upgrade or replace them is more difficult.
It's a strange phenomena.
I worked at a place that named all the servers after fantasy places: Landover, Andor, Krynn, etc. We had an SGI R3000 named Gastropod because it was so slow. My current Linux server is called skynet, though it's more likely to break down in some horrible way than take over the world. Lately I've gotten lazy - my MacBook is just called MacBook, and my two test Linux machines are linux32 and linux64.
15 years later though and my primary workstation is still called Andor.
I work for a research group in ecology and biology. I am terrible at thinking up names, but I wanted to be more creative than just using ID numbers. So one day I asked some cow-orkers to help out. They came back to me an hour later with a long list of genus/species names that I've been using for the last 12 years. Things like "Abies" (evergreen fir tree) to Vitis (grape vine). It was interesting to see that people became biased based on machine name. I can understand that "etheostoma" would be unpopular simply because it takes longer to type on an SSH command line, and is prone to misspelling. But people also didn't want the name "tortoise", because that somehow implied that the machine, which was identical to every other machine, would somehow be slower than the rest just because of it's name. (I also had one guy who didn't like the name "pinus" (pine tree) only because with a little mispronunciation it could be a phallic reference.)
Naming convention is Division of the company - Business Unit Location - Type - Unique ##:
DS-VT-WS01 (workstation) or DS-VT-LT01 (laptop)
Have a spreadsheet with the user tied to each workstation (changes aren't made often, but easy), also a Visio diagram of the building layout with system locations. This takes a little overhead, but stays up to date quite easily after the initial work load.
Cthulhu isn't a god, he's a great old one who serves as a high priest to the outer gods like Azathoth and Nyarlathotep.
Where I work we use the following convention:
Institution-Departmentxx
Institution can be a one letter marker, let's say the company you work for is DataTech, call it D.
Department is a shortened version of Payroll, HR, Accounting, IT, Shipping/Receiving. For example: D-ACCT, D-IT
The xx is a serial number of the machine deployed in that department: D-ACCT12 or D-IT58
If you want to retain information such as speed in a simple hostname, use the purchase year, such as: D-IT-09-58.
Simple, intuitive hostnames that still fit MS's constraints.
All of our workstations get named EXT000.domain.local (obviously 000 changes to the extension)...it makes easy to setup remote log-in via the VPN and it makes it easy to identify who has what installed on their computer uniquely. We're a small company (10 employees), so it may not scale well, but if you're really hung up on keeping track of individual computers, it seems like a no brainer to treat it like the telephone (just another tool) and tie it to that...
Where I worked, whenever you were upgraded to the newest, most powerful machine, there was some incentive to name your machine in such a way as to discourage others from creeping on and stealing cycles.
Slow sounding names was one way. Names that were hard to spell was another.
I almost named my machine "camouflage"
I settled on "potato" (it *was* the Dan Quayle era, after all)
Here are the prominent ones http://www.alexanderband.dk/gud/diktator/d2/index.htm
First letter is the OS: W, L, M 2-4 are the first 3 letters of our site code (CHI for Chicago) 5 is the class of machine: L = laptop, W = Workstation, D = Desktop 6-10 are the Asset Tag of the computer The final is optional if it is a VM we put an identifier letter at the end. V So, a laptop, running Windows in Chicago, that is a VM would be something like WCHICL012345V
I've always kept it simple. A 3-4 letter prefix designating which site the computer is at, followed by the serial number of the computer. If you want to store more information, just create a simple database using the code as the ID.
I used to work in a business which included a wan of a few hundred (perhaps around 1000) computers.
The admin named the workstations with simple IDs, the first digit might indicate which of the 3 physical locations the computer was etc.. while the servers were named after transformers; Optimus, slag, etc..
this was helpful because whenever I or someone else needed to set up a workstation and needed to put in the email server or gateway or dns server or we needed to remember which fileserver had the install files for microsoft office it was easy to remember. at least it was for me, since I was also into transformers as a kid.
Our policy in our office is pretty simple. Same one we recommend to our customers. Initials of the company name and 2-5 digits for workstations depending on company size. Initials of the company name LT and 2-5 digits for laptops.
So examples for a smaller company are SD01 and SDLT01
In regards to unique machines, we typically give them a slightly different name. For example we have a machine that controls our TV displays at the front of our office. It would be named SDtvdispay1, another great example would be SDsalescounter1
In the description, we put the username of the user. We also have an excel spreadsheet that we use to track purchase date, where we keep serial number, warranty information as well as any other critical information about the machine. We got away from putting serial numbers in the machine name, as its much nicer to look for SD01 than it is iy6BUKOYBI6g espicially when you could have iy6BUKOYDI6g sitting right next to it physically.
I personally have always avoided naming workstations after the user, due to past pain. We had done this with a customer of mine, and due to a poor application if you changed the workstation name, you had to call the software vendor and have them add it to the application server. Not to mention reinstall the application on the workstation. It all started because the CFO's ex-wife used to work there too. And he used to get all pissed off when he would go into network and browse for the business system and see "Karen" (small company). Long story short, at that point i renamed 30 PC's to the above format so he would shut up and stop bugging me about people who used to work there still being named on their old computers and spent about 2 weeks doing it.
That's not simple in the sense that it's not fault-tolerant. How easy is it to say or hear "AB" instead of "AV", for example?
If, on the other hand, your name is (for example) types of fish, it's totally unambiguous -- there's only one way to spell "pike", "salmon" or "stingray" (one of the way-back-when naming schemes in my alma mater). Yes, you'll have to change topics every now and again, but if you change topic with every change of technology, you encode an extra layer of information. So all your windows XP boxes are fish and your Sparcstations are countries, for example.
The extra information is a freebie, and can be ignored by anyone who doesn't need to know, but damn straight the techy will immediately know that the guy asking to have Windows Media Player installed on Botswana is barking up the wrong tree.
HAL.
Got them moderator blues I blieve I walk out the do', With these mod-points I been gettin', I 'most never post no mo'
Workstations should not need to be accessed over the network so they should not need a friendly name.
The key word their being "should not" in the real world however, I really don't feel like going down three flights of stairs to a cubicle in the basement everytime someone accidently changes some stupid setting which I can easily remote in and fix in a second. In my office we have little turn over, and as such name desktops X-username, and laptops Z-username. That way, when "username" calls with a stupid question, all we need to know is if it's a lappy or a workstation remote in without needing to do all the legwork. Lazy? Maybe. Smart? Definitely. I can see people's point about having unique names that never change, but it really takes very little time to change a name, and all people get new machines through me so there is no danger of a mix up.
Do not argue with an idiot. He will drag you down to his level and beat you with experience.
Two initials of the employee and their extension.
SL8846
If the user has more than one machine
SL8846
SL8846_2
SL8846_3
If you have multiple location, give each location a letter code as well (T for the Texas office for example)
TSL8846
Or something similare. Save these in a spreadsheet for quick reference of course, but with a naming convention like this each workstation user is easily located.
No sig here...
Here you have a great source of random names:
http://www.ikea.com/ms/en_CA/virtual_catalogue/IKEA_Catalogue.html
RFC 1178 addresses this.
It's 11pm, do you know what your deamons are up to?
I worked as a research assistant while getting my master's. The faculty that I worked for was Indian and liked to name his computers after Indian dishes. His fellow Indians did likewise and his Chinese students named computers after Chinese dishes (mooncake, wonton).
I decided to name my workstation ostakaka.
http://www.cooks.com/rec/view/0,1813,148162-228196,00.html
Reverse-DNS lookups on incoming connections are very handy for diagnostic purposes.
Accessible PCs are also helpful for local printer situations (like non-workgroup cheque or inkjet proof printers) that are accessed from a third-party database server or such.
- Michael T. Babcock (Yes, I blog)
This may have been posted already, but I've found that if you have a short "asset" number on most desktops, naming them after that is usually a pretty safe course of action. At my last job every computer have a sticker with a 6 digit "asset" tag on it, and all of the desktops hostnames were p123456. Every employee in the company KNEW what you were talking about when you said the asset tag, and since we did a lot of remote access via VNC, getting the user to tell you their hostname so you could remote became trivial. The same number was also in our warranty database and our inventory database, and it NEVER changes.
AccountKiller
We only have about 45 workstations, so we keep it simple using a 2-letter/2-digit scheme where the two letters represent the general function/department (TS=Testing, PR=Programming, CS=Customer Service, etc.) and the 2 digits are just sequential starting at 01 and increasing as we buy new equipment. We do recycle digits if a machine is replaced because people use a lot of shortcuts around here that refer to the machinename.
That's all. I like the toy-name-scheme by MightyYar, though :)
One of the more successful naming schemes I have used for workstations is naming them after the telephone extension where they are located. (Works best when each workstation has an associated phone.)
My workstation to user look-up table was the telephone directory.
When a user left and was replaced, the telephone extension list was always updated, and thus my workstation to user database was too.
When a user moved locations, they usually took their telephone extension and computer with them.
Most users knew their phone extension and thus I could easily ask them that information and obtain the workstation they would be at without looking in the database....or I looked at Caller ID.
All PC's here have asset tags on them so we put that in the name
-
example: d630-123456
all pcs are dell so you know thats a dell d630 laptop with asset tag 123456
it just causes fun conversations..
I have to go, the "Cleavland Steamer" needs to be rebooted.
And that damned PC for sales research is named "Dirty Sanchez" for a reason.
Do not look at laser with remaining good eye.
I used to work at a university. Software was tied to where the workstation was used (labs vs faculty offices vs office staff). When machines moved, they were re-imaged with appropriate software and names for their new location. In that environment, using building name + room # + station number worked well for labs and general office staff. We used faculty name for faculty desktops and notebooks because these sometimes floated between their offices and research labs (i.e. jsmithpc1, jsmithpc2, jsmithnb). As far as actually tracking down a machine, the name gave you a good starting point. We also kept a database of MAC addresses. If something wasn't where it was supposed to be, I used SNMP and a simple PHP app to find what switch port they were on. Failing that, block the machine at the firewall -- they then had a tendency to find you... This wasn't the most entertaining solution, but it was pretty functional.
Once I was reading some guy who named servers after painkillers.
Later I worked at a company where it was claimed that the new server would "cure all our headaches". I jumped on that and suggested naming it "Advil".
So they did. And Motrin followed.
Then for some reason we migrated off to antidepressants; Paxil and Prozac. I think that the shift said a lot about our collective unconscious there....
Prefixed or suffixed with a D or L denoting laptop or desktop. IE Desktop's SN# is 53e89c1 Host name would be if you prfixed it. d53e89c1 This gives you unique names for each machine. Also easily crossed referenced in an asset database that can show location info and assigned user... etc.
The way a PC is named can be really useful. When we were replacing our 5000+ machines, upgrading to XP, and implementing a radical new app delivery mechanism, we decided on this scheme: two characters for the OS, year it was deployed, machine category, and asset number (xp04p11111). It has worked out pretty well. Just by glancing at the name you knew pretty much what to expect if there were any issues. i.e. 04 is going to be a pretty slow machine compared to 07 and 08. "p" is a desktop PC and "l" is a laptop. We didn't want to put any location information in since we knew that they move around over the years.
The scheme has worked out pretty well, except for the times when the name is mis-typed ("O" instead of "0").
For the sake of making things easier on our SMS admins and the field team, we use the Dell/Apple/HP serial or service tag as well, since the manufacturer can keep the specs and the purchase order info themselves.
We do this:
Brand Code is either D for Dell, A for Apple, H for HP, etc.
And VMs under them are:
VM
So right now, my box is CISD6XQDMJ5, but I'm writing on a VM called CISD6XQDMJ5VM04.
The beauty of this is that it lets the admins on SMS easily select departments by building queries that say:
for all machines that begin with "CIS", do this thing.
or
For all machines that the fourth character is "H", do this other thing.
and
for all machines ending in "VM??", do -NOT- do this thing, since it might be hardware-specific.
As for location and/or username, that stuff changes too rapidly to adhere to, if I know what -department- the box is in, I'll probably be able to find it, and the serial number leads back tot he model on the web site, so I can go to Psychology looking for an OptiPlex 270 that's acting-up.
"Sometimes, I think Trent just needs a cup of hot chocolate and a blankie." -Tori Amos on Nine Inch Nails
There is actually an RFC you can refer to for help on the difficult problem of naming computers:
http://www.faqs.org/rfcs/rfc1178.html
This is how the loudness war is killing music.
All our workstations have serial numbers burned into their BIOS. During imaging, we have a bash script grep a hwinfo dump for the serial number string and use the result to create the workstation name. The same script also uses the hwinfo dump to automatically determine which hardware image to load onto the machine.
Any other info we need to track is kept in a database, with the workstation name as the key.
If you want to get REALLY fancy, check out ZENworks Configuration Management (ZCM).
];)
Regards;
Each location uses 3 letter aberration for closest airport, that is what we do across about 12 sites. Washington DC - WAS New York - NYC Miami - MIA --- Then when your in a city like Nashville and the Code is BNA... you suck it up and keep it standard. If there is no airport near by or two sites close to one airport, you make the most logical unused code. Then for security reasons you can use the floor number or some type of identifier for areas. Then MAC address or partical mac. You don't want to have a person who is at the computer as the the hostname, for security reasons. WAS2b001c378f3E3C
We put an asset tag on every machine. Then we use the asset tag as the machine ID/computer name. Unique, simple and right in front of the user when they call in.
There's nothing to change if the machine moves around. That's all handled through the automated inventory tools. Need to know the OS? Look it up by asset tag.
We would have used the accounting fixed asset numbers but those aren't assigned until well after the machine is deployed.
Serial numbers are an idea but they're not exactly easy to get from the end user when you're asking questions for troubleshooting.
Our asset tag format is X000000 with X being alpha. This gives us plenty of tags and they're all bar coded for that eventuality of needing to do a physical inventory.
Nothing silly, nothing hard to understand. Nothing that needs to be changed as the machines move around. Which they do.
Slightly off topic are the server names. Prefix for prod/test, hyphen, region, hyphen, identifier. As in TS-XYZ-APP01 for a test machine housed in the XYZ region for Application. Simple, easy, etc.
Once you start dealing with thousands of PCs and hundreds of servers you stop trying to be cute and make things as easy as possible.
I have not yet seen anyone post this rfc document which would probably be most useful to this question.
RFC1178 - Choosing a name for your computer
http://www.faqs.org/rfcs/rfc1178.html
During elementary and secondary school, they were named by building#-room#-comp# so High School Room 118 computer 3 would be 07-118-03. This could easily be adapted to other things, but usually the point is to track the physical location of machines not who is using them, with the exception of laptops (which had the naming convention Teacher's last name-LAPTOP and were in a different windows workgroup)
... of around 50 odd machines we use this naming scheme - --- where client_abbreviation is 2 or 3 letter short form of the client's name (services company), env is the environment - DEV, SYSTEST, CAPACITY etc., app_name is the optional application installed on that machine (db for database, app for app server etc.) and xx is a 3 digit serial number assigned incrementally in the order the machine was setup (first db server gets 01, second 02 and so on.)
So we end up with names like disney-app-dev-001 etc. Has worked well so far.
And let the adults talk about their chosen subject.
If Google really cared they would fix Android Chrome to reflow text, instead of discriminating
We started with PC0001 and are now at PC7921. Someday we'll need to add another digit. This isn't really that hard a problem to solve.
the company i work for has 20-some buildings around the state so we name ours like this:
OS-Location-Department-Unique ID
(OK, that might be an overstatement--there _should_ be exactly two answers, though)
You can follow one of two conventions when naming workstations: Functional or whimsical.
Functional names are self-evident - room number, system type, city, division, role, etc. The fine details amount to figuring out what information is important to you, and squishing it into an RFC1178-acceptable name.
Whimsical names are the other convention. Pick a theme with enough namespace for growth, and go with it. One company I worked for used cartoon characters. Another one used astronomical entities. A friend has his machines all named after single malt scotch whiskies.
The key is to make the namespace large enough for long-term unrestrained growth. A mid-sized company may want to make sure that they have three digits for workstation ID numbers (for functional names), or a ridiculously large pool of names to choose from. Hitting a self-created namespace wall is ugly and embarassing.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
Planet of the Apes characters for desktops.
Snow White & the Seven Dwarves characters for servers.
It was, admittedly, a small office.
Some of the advice is common sense still (like, don't use embarrassing names or misspelled ones), but it's from a different time, when having more than two or three computers serving any given role meant you were a REALLY BIG SITE. And you had to be important to have a domain:
From a security stand point, you shouldn't have a naming convention.
It should be random with the key locked up someplace.
A convention is a pattern, and this allows people to ahve insight into your system
That said, if you go with a naming convention, be sure to pick a topic that is wide enough to support your infrastructer growth.
I mean, you could go with 'Wizard of OZ' but after about 20 machines your going to be hard pressed to find unique names.
The Kruger Dunning explains most post on
Besides the general burden of overly cryptic names, site-based names have to be the stupidest thing I have encountered. Unless the machines are cemented in place, guess what? They move to new locations. So, now you have a site-based name that is irrelevant. Of course, you can rename the machine for the new site, but in my experience that breaks a lot of the software already installed on the machine.
"No matter where you go, there you are." -- Buckaroo Banzai
Every home system I've owned--including the iPods that my wife and I have used--have had music-oriented names. Jazz, Blues, Funk, Hip-Hop, Bebop, Fusion, Rock. I think that's it. The hard drives on the Mac systems have had movie character names, including Marcellus, Jules and Vincent, to name a few.
The CB App. What's your 20?
We barcode our workstations. The computer name is the barcode number. If you remove the barcode sticker, we will hunt you down.
I'm actually quite surprised that in all of the replies here, no one has remembered the actual solution for keeping branch office, departement, floor, room number etc. with a machines name.
It's called a subdomain and it rather works somewhat like this: x.dept.city.company.tld, so you'd get e.g. 01.accounting.amsterdam.acme.com. You can let the workstation get its hostname and domain information through a rather useful system called DHCP. In most cases, this is already being used to configure IP addresses, and given these addresses can change, the odds are pretty high that you already HAVE a database of names which already HAS a key for each record.
You can make as many subdomains as you like, and you can even delegate them to the seperate branches, departements, your stepmother or whatever you like. This also makes the system work more distributed, like the Internet, so if one nameserver dies it can never take down the entire network. Cool, huh?
Ghod I HATED this argument at the IT department.
"We should obfuscate the machine's use so hackers won't see easy targets."
You have to be kidding me. Most of the attacks won't bother trying to decypher some elaborate naming scheme, or get HR records to find out who's the CFO -- they'll just carpet bomb the entire network, and exploit any vulnerabilities they find. Frankly, if you have a hacker who already has access to the HR records BEFORE they break in, you've got a bigger problem, like maybe an inside job.
Workstation -> username. So when someone's downloading streaming porn and it's clobbering your bandwidth, you know who it is immediately. When the workstation changes hands to a new user, you should re-image the machine anyways.
We had the "serial number, referenced to a database" method at three locations, and each time I'd find out that someone was rushed and didn't update the database (or updated the wrong database), and I'd have to spend an afternoon validating all the entries again. This only served to slow me down, and didn't slow down our break-ins at all (which were, by the way, autonomous viruses and worms, not humans who could comprehend hostnames no matter what info we put there).
However, I did find some value in not naming machines by their purpose -- we did have a virus breakout that looked for machines named 'mail','smtp' or 'mx' for possible spam relays.
("why so many virus problems in places where you work?" I hear you ask. "are you some kind of shit IT guy?" No, I'm an IT guy that deals with C*Os who feel they don't have to follow the rules, and I get punished when I impose the rules upon them, even if it's for ISO 9001 compliance. Makes me sick; welcome to Toronto.)
Assign them to users, and give them a name like..
User-Classification-City(3 characters)-Iteration
Classification goes from INFRARED to ULTRAVIOLET, but INFRAREDs don't get any PCs assigned to them.
Iteration starts at 1 and gets incremented by 1 every time the PC needs hardware servicing.
Say, you have a guy named Peter, and he's from Amsterdam..
Peter-R-AMS-1
His PC breaks down and needs servicing?
Peter-R-AMS-2
Someone decides to promote him?
Peter-O-AMS-2
He breaks it another 4 times?
Peter-O-AMS-6
He breaks it yet again?
FIRE THE BASTARD! Anyone that needs a PC that requires that much servicing should not be allowed to access Friend Computer property!
Previous IT job was at a university, and over a decade, who was in charge of the naming conventions changed a few times. Each had their own idea of what the convention should be; department+num, building+num, department+type+num, building+type+num (type being mac or pc, when we were a mixed system). So say you have a building named "DM" with a dep't name "HPER". Right down the same hallway, I found;
dm01 ...
dmpc01
hperpc01
hpermac01
It was a horror show. Where I'm at now does a building+room/dept+type+num, so the 1st computer in room 30 of the high school is;
HS030C01
while the art room's computers in the middle school are;
MSARTC01, MSARTC02, etc...
Seems to work out ok.
I use the same method: [LOC1][LOC2][UniqueID]
LOC2 may not be needed, it depends on asset ownership. If the machine is a corporate asset and does not belong to a specific group/department/site then it is a corp code. The UniqueID is either the service tag, if present, or the last 6-10 characters of the serial number (depending on # of assets and variety of machines.) If a specific cost center owns a machine it is very unlikely that machines will move from that department then I add LOC2 to tell me better where the macine is located quickly. I do not care who is at the machine, as it makes little to no difference to me.
BGInfo will give the user what they need to tell me the machine, and I keep the information in a DB so I can reference it without interaction.
My method works for me and I do not have to change it when a user change is made or if the machine moves cubicles or specific locations, assuming I have done my job correctly, the only time it is changed is if a location or department is downsized or moved to another campus/address. Major moves are rare and most of the time include a reimage at which point the machine record in inventory is marked as "archived, not currently active" and the licensing is returned to the pool, the new image process takes care of the new record generation as if it is a new asset.
My previous experience in this matter was with a smallish multinational corporation with 20 or so offices around the world. It seemed to work fine for them.
Allowing business units to have their own assets isn't really a problem, but allowing that ownership to dictate IT processes isn't a great idea. Support should be standardized wherever possible to provide uniform service.
You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
(This is mostly related to WW2 naval naming schemes)
Some classes of British ships used the 'themes' (e.g., Black Swan-class sloops all named after birds, and first letter schemes (e.g. B-class destroyers (Bulldog, Boreas, Brazen etc). I believe these were most consistent among the smaller ships.
American ships: battleships (these days, nuclear subs) named after states, cruisers after cities, destroyers and destroyer escorts after people, submarines after fish. Other classes had other schemes, but as far as I know, none were purely alphabetical.
RoomInBuildingPCx The easiest way to do a naming convention is this. x represents any number from 1-? however many PC's there are in the room if the building has no name use a name that is common among your employees.
Workstation support is standardized. OS (lets sdaw win 2K3 support) is also standardized. Knowing the owner allows us to bill the appropriate party without having to look it up. Knowing the OS ensures a tech trained in 2003 server gets the ticket...
The problem is, we have hardware support, OS support, application support, DBAs, Web admins, SOA support, DR support, we have specialist for nearly everything, and understanding who owns what PIECE of a server (a lot of stuff here is middleware, or shared infrastructure as well) is important, and we don't want to have some massive database all 900 or so of the people with rights to access various servers has to have access to. We have a simple system of who "owns" (as in responsibility) each app, and that person gets a call anytime anyone elses serevr equipped with that app has an issue. By using the name, and without referencing AD or a seperate database to tablle, a tech can quickly reference the ownership.
But its not even techs... We have hundreds of scripts that run in response to certain errors from certain monitored systems, and the servername itself in the error (or the IP cross checkes in nslookup) provides key information to the script without having to JDBC enable it as well... When we get a nonresponse from a servername, the hardware owner and Os owner can be instantly notifed by output from our monitoring software to check it out.
Simply put, the servername structure, being consistant and containing a lot of data, provides simple automation of alert services, not just awareness for the engineers and techs.
There is no contest in life for which the unprepared have the advantage.
In Soviet Russia, workstation names you!
At my company we do site abbreviation and the serial number of the machine for the workstation. This is because all warranty items are tied to the serial number and you don't want to have to pull the ,machine out to fine out what its serial number is. .
Oh, wait. . .
What?
Using asset tags works fine until you get some special machine that was built up out of a dozen purchase orders and has twenty asset tags as a result!
The worst I ever saw was a $20K lab computer that was built up out of $500 chunks because the various managers could sign for that much without further approval. There were several boards that has five or six asset tags!
Worse yet the entire department was laid off a month after the thing was put together!
When I left the company there were still raging debates about how to classify that machine and which department actually owned it!
We name each system after their asset tag. We then have a database that records details such as serial, warranty status etc, all linked by the asset tag number.
Before systems were named after site_room_function, and that worked, but yes, equipment moves, so it's not very flexible.
Naming after asset tag has worked well for us - just keep a database with those details or it's useless.
That's a lot of work when someone changes a cubicle.
One place I worked had extremely movable walls even for offices. You could stay at the same location with reference to the building itself and yet change office numbers three times a year. When the company's fortunes were looking up, your office kept getting smaller; therefore, getting a larger office was not necessarily a good sign.
I echo the many recommendations above to just use an ID number and have a spreadsheet with the other data you need. It isn't sexy, but it gets the job done.
I'd also have a digit in the ID be a check digit (like in ISBNs). That way if someone typos a machine ID you probably won't find a line for it in the sheet (rather than finding the wrong line), and you'll know to ask again.
RFC 1178: Choosing a Name for Your Computer, http://www.ietf.org/rfc/rfc1178.txt
RFC 2100: The Naming of Hosts, http://tools.ietf.org/html/rfc2100
At my uni where there are ofc several hundred/(300-400?) computers they use a positional dependant name: ground level is 0, first floor is 1, etc, then there is the class room, followed by an enumeration of the computers in the class.. something like: //lvl0-cl08-pc1 would be in the class room cl08 on the ground level and classrom id: pc1. The servers are also in the same spirit: //lvl0-cl08-srv3
For the networks ive set up i always choose a theme and give them names accordingly: the solar system (jupiter is the big one, venus is that sexy lil macpro etc ;) :P
Think also all your mytheology if you want or something just fun
Just use major hurricane names .....
Where I used to work, we held a "contest" to determine what our naming convention would be. The winners were appliance manufacturers for servers (sunbeam, hotpoint, maytag, etc.), candy bars for workstations (reeses, heath, skor, crunch, etc.).
Of course after using all the common ones and digging into the past (bighunk, marathon, etc.) we eventually ran low on candy bar names, and had to "cheat" a little, using wrigley, twizzler, etc.
It worked well - we had fairly low turnover so the names became identified with the user over time. It seemed to give the servers and workstations a "personality" as well.
Absurdity: A statement or belief manifestly inconsistent with one's own opinion. -- Ambrose Bierce
Works great in NYC - and helps characterize the quality of the workstation (guess which boxes get names like "Flatbush" and "Atlantic").
Corporate Customers and systematic naming are fun.
International Shipping System: the ISS Cluster.
Of course it needed a development environment: DISS.
Then for HA testing you need testing cluster: TISS.
Then they rolled it out to production...
Discussing major outages with the suits was fun.
Seriously, the whole point of DNS was to provide a name that humans can remember. Ask to me remember "bob" or "elephant" or "pilsner" and I can tell you exactly where that machine is and what it does. I know you're asking about workstations, but here's a printer and its server that I actually use at work (XX denotes the company initials so that I dont get in trouble with the Man) 5C106HPCLJ4650 on XXsny195prt2. I think it would be easier to remember a dotted quad...put the info in a spreadsheet like others have suggested and give the workstations names people can remember and use. Maybe even have a little fun with.
Ever hear of HINFO or TXT records?
I run a tiny CAD lab. 4 computers in 4 cubicles, all aranged in a nice line. Orignally they were all SGI Octanes, so the names were pretty simple. Octane1, Octane2, Octane3, & Octane4, based on which cubicle they sat in. The number of times I was asked "Where's Octane2?" (or any other #) was downright staggering.
They're now all named "red" or "blue" or "green", and the doors have been spraypainted to match. Not terribly attractive, but one less stupid question to answer.
I used to work at the Democratic Party of Oregon. we just named each machine after a county in Oregon. When we ran out of counties we started using towns. The names were easy to remember (Malheur, Douglas, Wallowa, etc. as opposed to random numbers), so when we asked someone which machine they were having a problem with they usually remembered the name. We then just had an excel spreadsheet with name, make/model, location, specs, serial, software installed, license keys, etc. I hate numeric names because they're just hard to remember and awkward to work with. Pronounceable names, preferably ones that people have heard before, work great. Geographic names, names of famous people, names out of fiction (I'm typing this comment on Voyager), etc. all work great.
I might be stupid, but that's a risk we're going to have to take.
If your users are so lazy that they can't read the asset tag off the front of a machine you got bigger problems on your hand.
Remote desktoping AT102024.int.example.org is not significantly more difficult than moniker.int.example.org.
Most of the time I find this easier than trying to spell their names over the phone.
IMHO, machine names make poor databases.
Name your hosts whatever you want and keep information about the detail somewhere else. There's just too much info to be adequately conveyed inside of a name.
This always allows for better name choices. For example, drug names.
"John, are you on cocaine?"
"Nope, I'm on LSD right now."
"Well, then who's on heroin?"
"Nobody, I'm pretty sure."
"Wait a minute, I'm using heroin right now, wait until I get off."
and so on...
Where i once worked, because the tech's machine was always swamped with viruses i simply called it "incubator" ;o)
Hi,
I meant to say that they should not need to be accessed /regularly/ over the network so a friendly name is really not justified. If absolutely required we can always create a friendly alias pointing to the asset tag. Although if I were considering that I'd also be considering moving whatever service or resource to real server.
For example we have AT101245. This machine has an A record pointing to it's DHCP assigned address like AT101245.int.example.org.
The appropriate PTR record is also assigned.
We just use the type and serialize the numbers, usually LT-01, LT-02, LT-03... for laptops , WS-01, WS-02, WS-03...for workstations. If they are a multi-site client we would prefix the name with the location, Caloundra=CAL-WS-01....Our clients are all SMB and rarely have more than 100 stations. If they did, well...prefix a zero and so on and so forth. Naming by user or location within the building is usually not a good idea as these things change often. good luck.
the system that we use a work is pretty straightforward, every site has a 3 letter code, like new york is (NYC) then each computer is named after is dell service tag, so a PC might be named NYC559J31S the service code is unique and the location code tells us where the PC is in the business.
then we can use the serial to compare with a spreadsheet or go back to dell for specs, warranty, etc.
The downside of using servicetags as hostnames is that most service tags are printed on the back of the desktops, or in the case of laptops on the bottom.
This means that its not very convenient for a novice user to find, let alone read it aloud to a helpdesk employee while lifting the laptop with one hand.
I do recommend using servicetags as hostnames . Just put a sticker with the service tag on the front of the computer. Near the on button is recommended.
PS Unfortunately at my current job they have put in the year of purchase in the hostname.. e.g. PC-8098 is purchased in 2008. This means ofcourse that in 2010 we have our own little version of the 1999 bug, cmdb wise. PC-0002 could be 1990 or 2010.
I've worked in a few IT shops that offshored UNIX administrations. The server hostnames are now the names of Indian celebrities or landmarks. Just saying.
I still have an old Sun pizzabox in the closet I can't bring myself to rename.. Sparky.
-Troll, Flamebait, and Offtopic are NOT equivalent to disagreement.
At my company we use Region,Site Code,Serial/Asset #
AMRNYC123456
Not to mention things like remote-control sessions for helpdesk purposes, and assigning images for "on next reboot", and associating applications for installation, and so forth. Being able to easily tell from the name which machines are which can be very helpful indeed.
I don't advocate naming the computer after the user, though - at least not when there's a more useful "location" to provide. For laptops, which get carried around from place to place, sometimes the name of the assigned user is the most helpful location information you can be certain of.
-- The Wanderer
I see a lot of comments from people emphatically recommending against identifying the owner in the machine name. Since you're talking about workstations, though, I have to disagree. A user's workstation should be wiped and re-imaged when it's reassigned, so the issue of a machine changing hands and keeping the original name should be moot. Upon a machine's first boot-up, it should be assigned a new name.
If you want to track the hardware, use a standard numeric asset tag.
If you love process and standards, a "username-pc" or "username-laptop" standard seems perfectly reasonable to me. Personally, I see no reason why you can't let users choose their own machine names (so long as it's unique and machines are placed in a separate subdomain from your services).
Obviously, the rule for servers should be different, since those outlive their owners, and RFC1178 is a great place to start there. But for workstations? They should come and go (even if the hardware is reused), so who cares?
Sure, you can name you machines like that, but what if the person reading them doesn't have a futhark font loaded :-)
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Dude, in case the wrongness of putting SSN in the machine name didn't tell you it was a joke, the poster has the Day of Week in there and hits you over the head with a 2x4 about having to rename the machine every day...
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
In a typical corporate environment, your machines _are_ going to get moved this year or next year. Either your department gets more people, and moves to a different floor, or you lose people and the Real Estate Mavens consolidate your desks, or Alice leaves and Bob gets her machine, or Carol gets a newer faster machine and Dave gets her old one, or your startup gets bought and you move into your New Corporate Overlords' building.
Forcing you to rename machines when that happens is annoyingly disruptive; not renaming the machines when you move will rapidly start to annoy your IT people.
And of course, if any of your people have laptops (like _all_ of your sales people and field engineers) forget naming them by room number.
Naming machines after the users has some of the same drawbacks, but depending on your environment it's less likely to get you in trouble. If Carol gets the newer faster machine, she'll probably move her files to it, and you can rename the machine when you give it to Dave. But of course you might end up with "carolpc" and "carolpc2" for a while...
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
"Up is down today" "Charm is acting as Strange as last week"
Back during the pre-internet UUCP days, and even for a while after that, we had lots of series of machine names - mountains, beers, dwarves, dwarfs (i.e. Tolkien vs. Disney), composers. My wife's testing department named machines after psychoactive substances, anything from Ritalin to speed to Prozac to coke. At one point I was considering naming machines after common appliances (toaster, mrcoffee, xerox, etc.), but aside from the Trademark Police, that became less practical once some of those companies started networking their hardware.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Ok, you probably need some kind of reverse-DNS tag or MS Domain tag if you're using those, but what function are you trying to achieve with the names? Debugging? Boring names are fine. It's especially strange if your users have laptops, so the workstations are moving around, or if your client machines are virtual.
Naming servers makes sense, though once you've got enough of those, the names start to get boring (my print server is mo3980; it's somewhere in Missouri...) If you've got roughly one client machine per user, give them a user-centric name (mine's usually either bstewart or my email id, depending on which of our corporate underlords is running the desktop support this year.) If you've got lots of machines per user, whether real or virtual, make them username1, username2, etc. or else give the user a subdomain and let them name them, so it's fnord.username.engineering.example.com.
Naming machines in a shared lab sometimes makes sense, if different people need a bunch of clients at different times. My current network lab has some routers named after cities, some after baseball teams, etc., but mostly they don't have interesting names.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
location code - computer type - unique ID (last 4 or 5 characters of the serial number is usually a safe bet)
e.g.
WLGD1234 = Wellington, Desktop, unique ID
SYDL2345 = Sydney, Laptop, unique ID
You can also extend the location information if you have multiple sites (or rooms on a campus, for example) so that you'd have something like
MELC1R7D3456 = Melbourne, Campus 1, Room 7, Desktop, unique ID.
I worked at another (outsourcing) company that followed a very similar scheme, but they added the customer code at the front and one additional character to let you know which OS was installed e.g.
CCAKLLM1234 = Customer Code, Auckland, Laptop, Mac, unique ID
CCAKLLL1234 = Customer Code, Auckland, Laptop, Linux, unique ID
Ok, test time, what does this translate to:
CCSYDC1R8DW4242
Ultimately you want to keep it under 16 digits, I've seen longer names make management systems, directories etc throw tantrums.
Final point: location code doesn't cover roaming users that well, so you can either go for the location they're primarily based out of, or you can go with something generic like "INT" or "EMEA" / "APAC", or "NH" / "SH" (north/south hemisphere)
I name mine after curries vindaloo, roganjosh etc. And when I run out I Google troll for more curries then I have to go & try them out! Works for me.
I have seen a fuckton of different naming conventions due to working in outsourced callcenter for corporations.
I have come to the conclusion that KISS is the best method here. Keep it simple, stupid.
Location-type-designation.
Huey, Dewey, and Louie.
But you'll probably lose one during a ring plane crossing.
Hurricane Name list. I don't think anyone suggested it.
Garry AKA -Phoenix- Rising Above the Flames
Si hoc legere scis nimium eruditionis habes
ironically, it is up and down all the time. it attained this attribute after it received the name, not before.
Ask Me About... The 80's!
For time-served advice on this matter, see RFC-2100 (ha-ha-but-serious) from 1997, RFC-1178 (serious-but-ha-ha) from 1990, and RFC-1034 (just plain tedious) from 1987. You will note that these date back to the dissemination of the current DNS system ; as the system hasn't changed significantly, the advice on naming conventions is similarly unlikely to change.
From a more practical point of view ... look at the likely size of your domain ; apply an appropriate number of doublings (to account for unexpected growth, everyone and their dog getting both print and fax servers, then not releasing the names when the servers go away, another unexpected doubling) ; go and find a "jargon" dictionary somewhat bigger than you suppose you'll need ; start to use it.
Examples I've seen :
- departmental computers in a university department (not the lab computers) ; likely namespace size a few dozen ; naming theme "malt whiskies", with some 300 possibilities (and you can use blended whiskies for degenerate bastard machines that won't be around for long, or a sub-department.
- servers in a small company, for mail, proxy, mass storage, etc ; likely name space a handful ; theme was fictional servants, with dozens of possibilities before needing to do research.
- me? At home I use raw IP addresses. When the wife or daughter knows what an IP address is compared to a host name, then I'll worry about it.
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
At the research institution i work at we name machines based on a particular theme organized by their purpose. Most of the sequencing robot controllers are named after composers, laptops named after the x-men, lab workstations are given good jewish names, office workstations are named after their users' mothers, tech staffs computers are named after james-bond villains.
It does help a lot though with memorization of whose computer is whose and what it does. It's also useful for gauging how important a particular problem is based on what kind of name the computer has. For example a problem with a computer named after a composer is significantly more important than a problem with a computer named after someone's mother. This all comes into play in our ticket system when the person submitting a problem gives only a vague problem description and the computers name.
We used names of characters from the Rocky and Bullwinkle cartoon for servers.
BuildingName_Floor#_Room#_Name
EastHall_2_E213_Dave
Tech_0_CR9_Charlotte
I make the names alphabetically clockwise from when i enter the room, so you have a rough idea where each name is.
We're lucky we have web browsers these days, otherwise people would still want to try to turn their spreadsheet into an operating system.
At my job, we have a mix of department shorthand+username or dept shorthand+position/number. The main IT group is getting ready to implement serial number based naming conventions for the desktops and laptops. We are a primarily Dell office so that would mean the ServiceTag, which is short, but it can still be confusing and a little too much. However, I guess it helps ID a computer without even having to turn it on. I don't know if they are extending this though to network printers, servers (named based off of streets in our city), etc. I prefer a department in shorthand and ID# my self.
1. I don't like alphabet soup for names.
# ssh gla-1127394
is too darned awkward. And having a user read off soup over the phone is error prone.
Good naming conventions:
A: Give some clue as to what you are dealing with. At one job I had 11 flavours of unix, 3 versions of winsnooze and 2 versions of Macs.
B: Are pronouceable, not requiring spelling out over the phone.
C: Are not words in common use, but are recognizable.
D: Are not people's names.
E: Are more than a single syllable.
So the SGI lab on 4th floor were all birds -- bluejay, siskin... The secretaries' pool of windows boxes were all named after ancient goddesses. The linux boxes in grad offices were norse gods and heroes. Most of the individual profs workstations were named after their favorite famous mathematiction. (Hilbert wasn't bad but 'ssh chandrasekhar' was a bit much. chand was a cname really fast)
This way grad students would poke their head in my office and say, "Tanager's disk is whining, and it won't boot." and I would know right away that it was an SGI in the stats lab, and not an linux box in that grad's office. Further, it was far more informative than SL-121 as 121 may have actually been 112.
As to naming/licensing issues. There is merit in a computer having an asset tag, or id, whatever, which you use to link all the other info together. The name then becomes what you call it. If the role changes in a major way, you can change the name without screwing up the rest of the data associated with it.
With people, in the U.S. you do this with your SocSec number. That's your asset ID. But no one calls you 619-68-6160, they call you Mike.
When I was with this department I had a flat file that contained stanzas of the form:
Name ...
Asset: string
IP: string
CNames: string; string; string
MAC: string;
Room: string
PrimaryUser: string
OSver: string;
A script using this would rewrite my DNS files, my DHCPD.conf file, my YP files, That way it was all in one place.
In general whimsical names are used by people who have a very small network to maintain. Themed names are used by people who have a few hundred machines. Soup is used by groups of people who have thousands of machines.
Soup comes into play when you have an IT mob, instead of 1 or two guys. Soup happens when there is no association of particular machines oddities for the IT guys. Soup happens when you have a zillion almost interchangable units. -- 12,000 cisco routers, 8127 identical Dell Winsnooze boxes.
Soup happens when the memory of the IT guy is not capable of mapping the entirety in his own head.
When you work in a place that has nothing but alphabet soup, get out. There is/will be little freedom. You will be criticised for wearing shorts and sandals in the server room. They will start charging you for coffee, and will buy cheap coffee to boot. You will spend your time defending why you did trouble tickets out of order, and for every minute you spend working on a problem, you will spend 3 on documenting it and 6 more on filling out paperwork to get the parts to fix it.
Third Career: Tree Farmer Second Career: Computer Geek First Career: Teacher, Outdoor Instructor, Photographer.
It's often written on the outside of the case, every device on the subnet will be unique, lots of PC boxes and routers come with little peel and stick labels that already have the address, and it's consistent across all networks. It shows up in a database all by itself, allowing me to see when it was powered on, pushes data every three hours so I can see that it's still in the building. The first 2 octets give me the manufacturers name and I can read it with my rebuild script.
If you only have a handful of workstations in a home or small office, call them Starbug, ZORG-169, Kevin, and Meebo.
If you have thirty thousand workstations in a national or international WAN, then they can count themselves lucky if their name is anything more than their asset tag. And at least with the asset tag, you can tell a caller to read it off the side of the PC and then use it to remotely access the thing. Tracking's not an issue - just tell all servers to keep records of which workstation names and MAC addresses log in through which ports on what switches every morning, and you'll have a fair chance to be able to spot when a computer mysteriously changes offices or departments. If there's accurate records of which ports lead to which cubicles, you can even physically track them around the company (and if you have security cams in the ceiling in all locations, you can literally be looking over a caller's shoulder seconds later).
For a small office, I'll name workstations by function or location or some combination of both but higher level 'boss' PCs by their names or function.
For example: "acctnorth" for the Accounting machine at the north of the room instead of just "acct1", and "controller" instead of "bob".
- Michael T. Babcock (Yes, I blog)
Unfortunately that means that while I could connect to "accounting3" as a CNAME pointing to "AT101245", it still means that reading my logs shows a connection from "AT101245" which requires personal manual translation to correct, unless I start adding TXT records describing the present use of each machine ...
In either case, the CNAMEs or TXT or any other records will go unmaintained just like the original problem of unmaintained A and PTR records, so you may as well pick the one with the least headaches to your situation -- which for me means naming the machines legibly and maintaining those lists properly.
- Michael T. Babcock (Yes, I blog)
Integration with your asset management system could make these changes automatic.
I still think changing one CNAME record is a lot more trivial than changing the machine especially if your in a active directory/windows environment.
the other one's Harry.
When I was at GE Corp R&D, the names of the day were New York waterways. Thus the large fileservers were Hudson and Mohawk, smaller systems were Alplaus, Esopus, etc.
Individual groups used their own themes, such as fish, large (Whale), desktop (trout, halibut), and network connected devices (chum). After installing a group of servers for one particularly unpleasant manager, and being told that "naming is your job, not mine," I used the body fluids theme, and called them mucus, snot, vomit, and drool.
Why couldn't you do both? Give it a name, and also give it a serial or unique id. The name might change, but the asset database can translated that into the serial, which is what all of the other records are tied to. Along with a cross-indexed naming history of course.
All of the advantages of naming, all the advantages of unique id.
If your asset database can't deal with that, maybe it's time to ask them to upgrade their software before the next purchasing round.
Guessing there's a company policy against that...
Do what I do in my Civ4 games:
Capitol City name: 1
second city name: 2
third city name: 3
It's completely boring--uninformative, destroys individuality, lacks imagination, and is, therefore, perfect for any sort of workspace.
So easy...