Ask Slashdot: Becoming a Network Administrator? 480
J. L. Tympanum writes "After many years as a star programmer, I have taken a position which involves maintaining and rebuilding the in-house network of a small company. There are maybe 100 machines, a mix of blade servers running Linux and desktop PCs running Windows of all flavors. Basically, I have to learn networking from scratch. I have been given an 'unlimited' budget to buy routers, switches, etc., to set up my own little test network as part of the learning process. So the question is: what's the right strategy here? What routers or switches or other equipment should I acquire? What books should I read? Should I take classes from Cisco, Global Knowledge, my local community college, or somewhere else?"
Ignore Cisco (Score:3, Interesting)
Re:Step 1 (Score:4, Interesting)
What I find is that you'll start out with one plan, meticulously formulated through research and consultation and even after management has signed off on it... And then you'll find out that a half of the plan didn't make any sense or didn't in fact work the way those FAQs or sales people said it would, and the other half will be trounced by new demands from the departments you consulted because they neglected to tell you a part of their needs, or changed their minds, or read some article they read somewhere.
Here's what to do. (Score:5, Interesting)
I'm buried so far down here, I'm sure no one will read this. But here is what you need to do.
1. Before you begin, attend a Cisco / Global Knowledge CCNA bootcamp. You may not leave able to program routers like a master, but you'll learn how networks work.
2. Visit every PC, Server, Router, Switch. Put eyes on everything. Create a master spreadsheet. Document model numbers, IP addresses. Create Visio documentation of the way your network is set up. Document everything. You need a good deal of cabinets to store it all.
3. Decide what is the most deficient part of the network, fix it with the simplest solution. If you're using hubs, buy switches. If the routers need to be rebooted constantly, buy new routers. Above all, keep it simple. If possible, stay away from V-Lans, encryption software, Linux, or anything else complicated. Do this every year.
4. Buy one third of the total number of PCs of the network plus ten percent. Buy only one model. Create a central image with Acronis and modify that image as necessary. Deploy these models. Repeat for the next three years.
5. Outsource security. That way, when it breaks you can blame someone. At the same time, make sure you can monitor security to prevent breakage.
6. If possible, outsource your main application. You don't want to support the product that everyone in the institution depends on. You need to keep the network up, not software.
7. At the end of year one, bring in a network assessment. Tell the assessor what he needs to find before he arrives. Use that the next year to justify your new purchases.
8. Make sure you stay friends with the president / CEO. When it is necessary to reorganize the server, etc, it will be necessary to have his good will.
9. Be prepared to work like a sunuvabitch for two years. Take your spouse / GF out when you can.
10. Don't let them make you program again. You're a network admin. You cannot support your old programming team.
Re:Why? (Score:4, Interesting)
Small businesses tend to have rapidly-changing needs and few staff. If they have less development work coming in, and a pressing need to replace a sysadmin, it's perfectly sane to ask the developer if he can switch hats, given sufficient resources and support. For the employee, it keeps him in a job. For the company, it saves them from having to hire a new guy, which is neither cheap nor enjoyable, and they'd have to train the new guy anyway, which is freakin' hard when the senior sysadmin is already long gone.
I don't think it's such a stretch, the two roles tend to complement each other quite well. A good programmer-analyst already possesses 2/3rds of the knowledge required to be a competent sysadmin. You know the shell scripts will be a work of art :) I don't know why you think it's at the bottom of the ladder, because I see it the other way around. Programmers are a dime a dozen (see China). Good sysadmins are damn hard to find, which is why I have no shortage of contracts coming in from past employers and acquaintances. Trust is a big factor, because really, the sysadmin controls access to every resource, and thus by necessity has unlimited access to all your data and equipment. Who would you trust more, some kid walking in off the street with the price tag still hanging off his jacket, or an employee you've known for years ?
Re:Odd choices (Score:4, Interesting)
My story in a nutshell.
Hired to program. Soon after system admin leaves. Server's need patching, junior admin screws up some compiles, etc so I step in and fix the server environment. Congrats, you are not a system admin (doh). A few months later, network admin is gone as well. New network guy is hired, but sucks at his job and for some reason doesn't get fired (still can't figure that out). I need the SAN to function properly, and I need the network to function properly. Congrats again, you are a network admin.
Now the title outside my office says "Programmer", but I haven't written any programs in at least 2 years. I've wrote a dozen scripts to make my life easier, but mostly I spend my time managing, install, patching, supporting, and planning the network and server infrastructure. Somehow I've also managed not to screw it up and have finally gotten to a point where I think I might be good at this. But I miss my compiler....
Re:Step 1 (Score:3, Interesting)
When I started as an admin 5 years ago, the company didn't know to care about redundancy, or security. When I started, neither did I. I could build PCs, do some light programming, and had a knack for finding solutions with Google. In that time, I've replaced all network hardware and fixed the topology, expanded from 6 to 20 servers, added virtualization wherever possible, added battery backup to everything (many servers didn't have any UPS), replaced 100 windows 2000 desktops, added 100 more desktops, upgraded the domain from 2000 to 2008, Exchange upgrades twice, migrated all storage to redundant RAID on server or via NAS, maintained DAILY tape backups of all servers, network monitoring via free Linux tools, expanded the network via T1 to include 7 satellite facilities and WAPs with VPN/firewall, and locked down every damned machine so that nobody can install anything. All while providing these people and locations with 24/7 tech support and software instruction. Monitoring scripts are all Perl and PowerShell, depending on OS. All of our network hardware is either Adtran or 3Com (now HP) and I've only had one switch failure in 5 years. No training seminars or certification taken, just lots of reading.
It pays shit, but its steady.
FML.