My first experience as a net admin was dropped on me in an indirect way. I applied for a job where I was told that I would be looking after a stock of freestanding PC clones that were used for research. I accepted the job. As I was leaving, they mentioned—as an afterthought—that they were having a new server delivered next week. The building was being cabled and there would be a small network to set up. The rest, as they say, is history. I learned network administration the hard way, making mistakes as I went. By sharing a bit of what I’ve learned, I hope to keep you from making the same mistakes I made. So, I offer these seven pieces of advice for net admins who are just starting out:
- Label everything.
- Back up everything and advise others to do so.
- Plan, plan, and plan some more.
- Communicate the need for downtime.
- Discuss users’ requirements.
- Use all available resources to learn new skills and techniques.
- Test everything and question everything.
The first of two articles
This article will discuss the first four rules. Next week, I'll follow up with a discussion of rules five through seven.
Rule one: Label everything
It may seem rudimentary, but the following anecdotal conversation illustrates why it’s important to document and label everything:
"Hello? I can’t log on to the network. Can you help?"
The help desk checks the network and surprise! The workstation in question is nowhere to be seen. According to records, it is connected to port number 2/21/12 (floor 2, room 21, port 12).
"What port are you connected to?"
Aha! They moved the office around and didn’t mention it to the net-ops people.
"Okay, that system has been moved. We need to patch in that port and disconnect the old one. We’ll have it done in an hour."
Imagine how long it would take to track down the problem if everything weren’t documented correctly. My advice is to label every port, every cable, and every terminal and keep a spreadsheet of who has what and where it is. You don’t want to spend your valuable time tracking network problems caused by devices that are simply moved and not reconnected properly.
Rule two: Back up everything and advise others to do so
Before the network arrived at the business of one of my former employers, we used to visit each workstation with a tape streamer and make a copy once a month. Consequently, if a workstation hard disk failed on day 26, all hell broke loose. The entire month's output from that person would be lost.
It was obvious that the company had grown to the point where something more efficient had to be done.They were considering employing another person whose job it would be to back up desktop systems all day long, when it was suggested that we get a network and a server. It soon proved to be more economical and efficient to wire all the PCs together and use a central server that could be backed up as often as required. However, it was then up to individual users to back up their vital information on an as-needed basis. That’s why it’s essential for the net admin not only to back up everything but also to remind others—frequently—to back up everything. You should instigate a strict policy of backups and advise users that any work they want to have saved should go on a network drive.
It’s also important to test your backups. It’s useless to have backup systems that aren’t working properly. Do an occasional restoration on your test network to verify their condition.
Rule three: Plan, plan, and plan some more
Don’t let your network grow unchecked. Plan everything from the color of the fly leads you are going to use to the size of the patch panel and where it is going to live. When the first network install takes place, it is easy to follow. Inevitably, though, there will be amendments to the network. You may need to replace patch leads, and soon, if you do not maintain your original plan of labeling and color-coding, it will be hard to identify all the connections correctly.
Plan what will happen in the event of a failure and list all the kinds of failure you are likely to encounter. Are your hard disks hot swappable? Do you have a spare? These days, with the low price of components, having a couple of disks you can bolt into a tray and slide straight into a rack may mean the difference between no downtime and wrong-time downtime.
Rule four: Communicate the need for downtime
On the subject of downtime, there are bound to be occasions when it's needed—anything from having the server room repainted to fitting a new replacement server or patch panel. These outages need to be planned. Obviously, you don't want to stop the work of the company if it can be avoided. In the past, I have found it best to negotiate for downtime, so that the people you are working for realize that this is an occasional requirement.
In my first net admin role, everything ran smoothly for several months, but eventually I needed to perform an upgrade that would involve an hour's downtime. I asked when would be a good time to take the server down, and the response I got was quite shocking.
The directors were horrified that the server, in which they had invested all their hopes and dreams, was in need of attention. There surely must be something wrong! It took quite a lot of talking to convince them that I couldn't perform upgrades without downtime. Had I been more experienced, I would have talked about downtime a long time before I actually needed it.
More advice from gurus
What's your number-one piece of advice for a novice net admin? What’s the worst communication gaffe you experienced as a new administrator? Send us an e-mail and offer your advice to those new in the field.