IT management - scalable approach

For a long time he was a passive reader of Habr, but he gathered his courage and decided to share his experience. I hope my ideas will be useful in the daily practice of admins.

A couple of months ago, I conducted an analysis of IT infrastructure at work, during which it was necessary to identify the weakest points and make recommendations for improvement. I am sure that the situation I encountered is known to many. Over several years, the company has changed significantly and has noticeably grown, however, some elements in the used IT infrastructure have remained unchanged over the past 10 years. As a result, various patches and refinements were regularly invented in order to put everything together, which obviously did not add stability and efficiency to the overall system. The result was a review of the overall infrastructure and requirements of business processes in order to choose a new unified solution. In other words - revolution, not evolution.


')
In many, if not all, small offices, system administration begins in a rather wild way, without any plan for the future. This, in general, is not surprising, since management cares more about financial well-being, and a newly-minted administrator rarely thinks about his place of work after 5-10 years. However, if you are still “lucky enough” to work in the same company for so long, you may well find yourself in a situation where everything seems to work as intended, but the number of isolated and incompatible systems improperly exceeds the labor resources of the IT department. This is exactly what happened in the organization where I analyzed the IT infrastructure, which once began with one server and two IT workers, and now it has hundreds of user machines, its own data center and a high-performance mainframe.

The following recommendations are a brief excerpt from a general analysis document and provide tips on building IT systems for future development and potential deployment of new products.

Data management

• Decide in advance how you will store all the credentials (user names, machine configurations, warranty records, finances, and so on). Avoid notes, notes on a piece of paper, copies of checks and the like - all this will be lost sooner or later, no matter how hard you try.
• Choose a single platform to manage such data. It is better not to rely on a specific product (Microsoft Active Directory, Novell eDirectory, (Open) LDAP, etc.), if this is not a specialized package for IT management. Ultimately, requirements may go beyond the functionality of the product, so it is very important that there is a public and widely used access interface (for example, LDAP).
• Try to always integrate new systems into the chosen platform. The presence of three or more isolated systems (for example, a corporate website + mail, a Windows domain and accounting for Internet traffic) is a clear sign that it is time to think about the future of such an architecture.
• Try to keep everything there , including data on work performed, machine configuration, warranty service and software licenses. This will give you the opportunity to learn about the history of a particular station and its user, and will help prove that you are not an elephant; your actions are legitimate in case of checks. In addition, you will have something to show the authorities on the question of what you were busy.

Client approach to users

• Treat your users like customers. Of course, everyone has heard about accountants and cleaners who are inexperienced users, but ultimately it will help you to differentiate your area of ​​responsibility from the field of black magic tricks and dances around all electrical appliances.
• Keep a record of all user hits. You can remember employees perfectly well in person, but when the IT department grows into another employee, and the company increases several times, it will be impossible to remember and explain to each other who and what was demanded of you.
• Give users the ability to track the history and status of their calls , as well as send requests remotely. This will give you time to work instead of wandering from office to office or listening to problems on the phone.

IT tasks

• Automate everything that can be automated . Of course, it is always important to evaluate “whether the game is worth the candle”, but when you create any code you win doubly - speed up the solution of the existing problem and create a tool that may be useful in the future.
• Create a bulletin board (in material or electronic form) . The most annoying situation is when the main means of communication between employees falls (for example, chat or e-mail), and while you are busy restoring the service, they constantly call you asking “what happened?”
• If you are not the only IT employee, select a special person (one or more) to receive calls from users. It is always better when you can concentrate on your work and you are not distracted by the phone or “visitors”.

Knowledge Base

• Create and maintain a knowledge base (Knowledge Base) . Write there all the non-trivial solutions that are classified as “made and forgotten”. Also try to integrate it with the configuration of all stations. In this case, it will be easier for you and your colleagues to solve problems that have already been solved by someone in the past. In addition, if the apocalypse happens and you have to reinstall one of the key servers, such a knowledge base will significantly speed up the process.
• Create a list of the most frequent problems and solutions to them . These can be known system bugs or frequent user requests.
• Add the ability to automatically save system logs from the most critical services and keep these records for a long time. Thus, you can track what happened with the system recently that will help you understand the reason for the error.
• Give users partial access to the database, and the ability to add their comments and posts . Believe me, some of them may understand their field better than you. In addition, it will allow other users to try to solve the problem themselves, before calling you for help.

The most important

Always keep an emergency copy of all your data , especially if you implement something from the above. Such a copy should not be on a parallel server or on your personal machine, but preferably somewhere on the other side of the city, or, better, another continent. At any moment a situation can occur (blackout, fire, mask-show), when you stay either with completely killed equipment, or without it at all. In this case, it would be a shame to lose with him the very base that would help restore everything.

Of course, in all of these recommendations, it is important to choose a balance between the available time and the need for implementation. Much of this is available in ready-made products for IT management, which the director of IT departments of large enterprises know not by hearsay. Unfortunately, most of them are proprietary (Remedy, LANDesk) and cost a lot of money, that is, if you work in a small company, the probability of introducing such a product tends to zero. However, there are free open source solutions, often quite simple ( SimpleTicket , Liberium , osTicket ), but there are more functional ones, such as OTRS . Most of them, unfortunately, do not offer Russian translation. Yours truly is currently developing the SAAS version of IT management, but this is a completely different story;)

I did not write the article, I posted it at the request of the author. If you want to send an invite, the address is letoosh@letoosh.com