Setting up a Small Office Network? 88
A not-so-anonymous Anonymous Coward asks: "I am embarking on a startup with some business contacts. I'm the only tech-guy in the group so I'll be the one to set up the network for our small office of 5-7 people. I've spent the last 15 years immersed in the development end of things (numerical analysis software and parallel computing codes). The downside of this is that I'm quite naive when it comes to networking: there's always been someone else taking care of revision control, backups, security, servers, etc., even purchasing stuff and running cable. What advice would you give someone who isn't afraid to roll up his sleeves, but is starting from ground zero on setting up a small office network? Can you recommend any books that are up-to-date and practical (e.g. "howto")?"
Been there done that! (Score:5, Insightful)
Re:Practice Makes Perfect (Score:2, Insightful)
Make sure that you leave plenty of room for growth. Buy bigger than you need; for instance, start with a 24-port switch instead of a little 8-port. You don't need those extra outlets now, but wait a bit and if this business really takes off, you'll be glad you got started that way. Establish standards for your hardware... if everyone's PCs will be using the same motherboards, you only need maybe 2 spare mobos in stock.
If you just think ahead a bit, you'll solve most of the pesky little problems before they even exist. For the rest, you'll need to start learning as much as you can about networking... the "for Dummies" books are good, like someone said, and if you're doing Windows stuff, you might want to get the study materials they offer for their certifications.
Documentation (Score:4, Insightful)
Or, if you happen to leave, you're leaving a good legacy for the next guy.
I know documentation is the bane of everyones existance, but when you're designing a new network from the ground up (including servers, workstations, etc) a little pain == lots of gain.
Additionally, create (and document) some quick policies and procedures now - things like passwords/external access/storage locations/naming conventions
Good luck.
Re:Outsource the bunch (Score:4, Insightful)
Move your revision control, backups, security, servers, etc stuff to India.
Don't listen to him. Look, I can get you a much better deal by nearshoring your IT operations with a Canadian firm who will then offshore to a firm in China. The Chinese firm will outsource to a contract company in Vietnam. Your cost will be next to nothing.
Uncomfortable with the plan? Think of it this way, you will easily be able to access^Wbuy back your business processes, security, and data at any time through a multitude of interesting and multilingual extortionists. Your company's operations will be mirrored across teh internets in a distributed network of black and not-so-black markets. You'll always have access to your company's IT operations regardless of how bad you mess things up in the USA.
The myth of "grow or die". (Score:5, Insightful)
Or, one could assume that the people involved are interested in having and controlling a successfull small business. Sure, you may not get stupidly rich, but there can be a lot of satisfaction in simply controlling your own destiny and turning a profit every year.
And even if you *do* plan to grow the number of people in the company, the likelyhood is that you're not going to do it in the first six months, and that when you do, you're going to *move*. In the meantime, building up infrastructure is a waste of time. Building a small, reliable net that you don't have to dick around with on a daily basis should be your goal. When it's time to grow, buy new stuff. It will be cheaper and more capable then than it is now.
Re:I can offer some advice... (Score:3, Insightful)
Until you need to do things correctly. Just playing with it willl get the basics working, but it won't teach you anything about the best practice for any given situation. It won't teach you the correct way to devise a backup rotation. It won't teach you how to devise an efficient addressing scheme. It won't teach you how to properly secure your network and your data.
Depending on your level of experience, you may already know some of these things, but the OP made it clear that he doesn't. Telling him to ignore the work, advice and experience of the others who have gone before is irresponsible, not to mention likely to waste a lot of time that he probably doesn't have to spare.
Another valuable guide that hasn't been mentioned is from oreilley
Essential System Administration, Third Edition By Æleen Frisch Third Edition August 2002 ISBN: 0-596-00343-9 1176 pages, $54.95 US, $85.95 CA, £38.95 UK
this is oriented to the UNIX admin but the advice is relevant to all OS
Step One: realise you know nothing (Score:3, Insightful)
The various for dummies, for complete idiots, and similar books are generally good introductions. You are going to need more then one book. At a minimum one that deals with networks, and one that deals with system administration.
You will need to decide on a maximum budget for IT stuff. You can easily spent that ammount, regardless of how much it is. Like buying a car, it is good to determine you maximum ahead of time.
You will need to determine what services your network will provide. Just Internet? Will you want a network printer? A file server? Do you need a revision control depot? Do you want full backup of each client over the network? While working out these details remind yourself that you are building a business network, not a home network.
You will need to decide what OS you will use. A single OS is always easier to work with. A free *nix server will save you money. Your choice of OS must be requirements based. Don't choose MacOS because it is fun when you will need to run an application that is only published for Windows. Don't overlook the alternatives because "windows is the default."
Once you have a basic design you may want to look at outside help. Chances are you know a sysadmin that you would hire on for a short contract.
When putting the system together aim for the mid priced solution. The cheap stuff will cost you time and effort to maintain. The cadillac solution is expensive and you will probably want to replace it in a year or two anyway. The mid priced stuff is almost always going to work well enough. Avoid ordering too much through priceline. Building a good relationship with a local vendor is worth more then the few extra dollars you will pay. Alternatively the (business) tech support from IBM, HP, and Dell is good despite the complaints on the Internet. However IBM, HP, or Dell won't sit down to talk about your setup, while a local vendor will. (do shop around for vendors though. too many are cheap discount shops selling crap.)
Keep it simple (Score:2, Insightful)
Try and avoid overcomplicated solutions to problems. If people are pulling you in different directions, try and get them to talk it through together first.
If there's a modular way of doing something and a non-modular way, pick the modular way so that you can change one element of a solution in the future without throwing everything out.
Ask yourself who's going to be doing the ongoing maintenance. If it's you, also ask what sort of support you want to give (hours of coverage, that sort of thing). It's worth setting some sort of expectations up front (including "if you screw your PC up by doing exactly what I told you not to, don't expect to be at the front of the priority list).
Get the basics right - backups (including off-site), security of access to data both from outside your network and if appropriate from inside, and security patches on servers and desktops.
Reading books is useful, but there's a wealth of information available for free in public (especially on Usenet), and "learning by doing" is often the best way.
Have some sort of test systems that allow you to try stuff out first before doing it for real.
Think about the sort of "disasters" that might happen and try and plan for those that might. Planning doesn't have to be a detailed step-by-step list, but it is worth thinking about potential problems before they happen.
Sorry if this sounds like Mr Bleeding Obvious, but you'd be amazed how many people don't do this stuff. You've done the first (and most important) thing already, which is start asking questions.