Memo for the office sysadmin

I have not yet met system administrators who would have intentionally harmed the company. In all cases known to me, system administrators created problems for companies solely out of the best of intentions, trying in every possible way to please the current needs of the company while neglecting the long-term consequences of their decisions. This is especially true for beginner specialists, whose level of best intentions just goes off scale.

In this memo, I collected benchmarks for office system administrators who help to achieve high-quality information systems and build relationships with users and businesses. I hope that this memo will help novice system administrators to comprehend the experience and draw the right conclusions for the future.
')
Further briefly and without water:

1. A computer in the office is a working tool.

A computer is the same working tool as a hammer, saw, drill, tractor. The office computer should help employees work, not entertain them during office hours. On the computer in the office should be only those programs and applications that the employee needs for work.

2. User - a person without rights

Dozens or even hundreds of users daily try to accidentally break something in the systems you have built. The more rights they have - the more opportunities for this. By limiting user access to systems, applications, and information to the level they need to work, you reduce their chances of causing damage to the company.

3. If the company is not ready to buy it, then it does not need it.

The driver does not collect the Mercedes from the Zhiguli and does not steal it from the manufacturer - it drives and repairs the car that the company was able to buy. You also do not need to build a fault-tolerant cluster of old system units and illegally use programs that the company does not want to buy. If companies really need such solutions, she will buy them.

4. Users must be able to work with a computer.

Knowledge of PC, Word, Excel in vacancies is indicated for a reason - the user must know how to insert the cartridge into the printer, connect the keyboard to the computer, and how to use the applications necessary for the operation. The only exception is employees who have a personal assistant for these tasks.

5. Your work is more expensive than hardware and software.

Computers and programs work 5 or more years and during this period the cost of your salary will exceed the cost of their purchase. If you want to save companies money - buy those solutions that will not waste your time and work in the company as long as possible.

6. Once every 5-7 years, IT infrastructure will have to be completely updated.

Nothing is eternal. One day the equipment will become obsolete and begin to bring more problems than good. The only question is whether you can replace old equipment with new ones before it creates a headache for your company or not.

7. There are no economic miracles.

An army of marketers is constantly thinking how to make more money on those who want to buy cheaper. If the low cost of the solution is obvious, then, most likely, you are strongly led by the nose. The ability to read the total cost of ownership helps protect against the tricks of marketers.

8. Think 3 months saves 3 years of work

Planning is never redundant. Especially when it’s built now has to be serviced for five years or more. It is better to be punished once for disrupting the launch of the system, than to correct mistakes during several years and endure complaints about the quality of the work done once.

9. Made neatly better than done quickly.

An effective system administrator is not one who does one hundred tasks per hour, but one who brings the tasks to the end. It is the ability to concentrate on the task and bring it to the end eliminate the situation when you need to do a hundred tasks per hour.

10. Simplify, structure, standardize or die

When you have one server and ten users, you can do anything, as you wish, using any application and in any quantity. The problems will begin when the infrastructure grows to dozens of servers and hundreds of users, and then, instead of administering information systems, you will be busy administering a mess. Take care of your youth!

11. Everything can break

If it does not break by itself, it will be broken by someone, burned in a fire, filled with water, or simply disappear in the cover of night. Absolutely reliable hardware and software does not exist. Reliability is determined only by your willingness and company readiness for any situations.

12. Backup copies are valuable company information.

Backups are not only a way to restore systems in the event of a failure, but also the ability to access information on some date in the past. Determine the number of backups, the depth and reliability of their storage should be in conjunction with the management of the company.

13. Always check someone’s diagnosis.

The surest way to not solve a problem is to solve it on the basis of someone else’s diagnosis. It does not matter who told you this diagnosis - the user or your colleague. If they could not solve the problem on their own, then there is no reason to believe that their diagnosis is correct.

14. The more serious the accident, the longer you need to drink tea before eliminating it.

At the time of the accident, adrenaline rises to unprecedented heights. Attempts to do something in such a state often cause more damage than the failure itself. A mug of tea helps to recover and to plan actions for the diagnosis and elimination of the accident. In my experience, the time for drinking tea in such situations always pays off.

15. Do not trust the backup system.

It's a shame when a developer's error, your mistake in setting up a backup, or a failure in the repository lead to unreadable backups. It is doubly insulting when you need these exact backups desperately and right now. To avoid such situations, regularly check backups manually.

16. The best sleeping pills - notice that everything works

Monitoring systems usually send messages only when a failure has already occurred. The morning notification that backups were made successfully, key nodes in the network are available and disk arrays normally extend sleep by 30 minutes. Trifle, but it took me ten sleepless years to reach it :)

17. Nobody wants you to flinch from a phone call.

Everyone wants the systems to work as expected. The problem is that apart from you, often, no one knows how to achieve this. It is your task to shape the expectations of users and company management, as well as designate the necessary resources and conditions for their achievement. If you flinch from a phone call, it means that you did not do it or did not do it to the end.

Successes!

Ivan Kormachev
Company "IT Department"
www.depit.ru