1.6. Tasks of the administrator

As we have described, we could divide the tasks of a GNU/Linux administrator (or UNIX in general) [Lev02] into two main parts: system administration and network administration. In the following points we will show in summary what these tasks in general consist of for GNU/Linux (or UNIX) systems; most part of the content of this course manual will be treated in a certain amount of detail; most of these administration tasks will be developed in this course manual; for reasons of space or complexity, other parts of the tasks will be explained superficially.

Administration tasks encompass a series of techniques and knowledge, of which this manual only reflects the tip of the iceberg; in any case, the bibliography attached to each unit offers references to expand on those subjects. As we will see, there is an extensive bibliography for almost every point that is treated.

Local system administration tasks (in no specific order)

• Switching the system on and off: any UNIX-based system has configurable switching on and off systems so that we can configure what services are offered when the machine switches on and when they need to be switched off, so that we can program the system to switch off for maintenance.

• Users and groups management: giving space to users is one of the main tasks of any systems administrator. We will need to decide what users will be able to access the system, how, and with what permissions; and to establish communities through the groups. A special case concerns system users, pseudousers dedicated to system tasks.

• Management of the system's resources: what we offer, how we offer it and to whom we give access.

• Management of the file system: the computer may have different resources for storing data and devices (diskettes, hard disks, optical disk drives etc.) with different file access systems. They may be permanent or removable or temporary, which will mean having to model and manage the process of installing and uninstalling the file systems offered by related disks or devices.

• System quotas: any shared resource will have to be administered, and depending on the number of users, a quota system will need to be established in order to avoid an abuse of the resources on the part of users or to distinguish different classes (or groups) of users according to greater or lesser use of the resources. Quota systems for disk space or printing or CPU use are common (used computing time).

• System security: local security, about protecting resources against undue use or unauthorised access to system data or to other users or groups data.

• System backup and restore: (based on the importance of the data) periodic policies need to be established for making backup copies of the systems. Backup periods need to be established in order to safeguard our data against system failures (or external factors) that could cause data to become lost or corrupted.

• Automation of routine tasks: many routine administration tasks or tasks associated to daily use of the machine can be automated easily, due to their simplicity (and therefore, due to the ease of repeating them) as well as their timing, which means that they need to be repeated at specific intervals. These automations tend to be achieved either through programming in an interpreted language of the script type (shells, Perl etc.), or by inclusion in scheduling systems (crontab, at...).

• Printing and queue management: UNIX systems can be used as printing systems to control one or more printers connected to the system, as well as to manage the work queues that users or applications may send to them.

• Modem and terminals management. These devices are common in environments that are not connected to a local network or to broadband:

  • Modems make it possible to connect to a network through an intermediary (the ISP or access provider) or to our system from outside, by telephone access from any point of the telephone network.

  • In the case of terminals, before the introduction of networks it was common for the UNIX machine to be the central computing element, with a series of dumb terminals that were used merely to visualise information or to allow information to be entered using external keyboards; these tended to be series or parallel type terminals. Nowadays, they are still common in industrial environments and our GNU/Linux desktop system has a special feature: the virtual text terminals accessed using the Alt+Fxx keys.

• System accounting (or log): to check that our system is functioning correctly, we need to enforce log policies to inform us of potential failures of the system or performance of an application, service or hardware resource. Or to summarise spent resources, system uses or productivity in the form of a report.

• System performance tunning: system tuning techniques for an established purpose. Frequently, a system is designed for a specific job and we can verify that it is functioning correctly (using logs, for example), in order to check its parameters and adapt them to the expected service.

• System tailoring: kernel reconfiguration. In GNU/Linux, for example, the kernels are highly configurable, according to the features we wish to include and the type of devices we have or hope to have on our machine, in addition to the parameters that affect the system's performance or are obtained by the applications.

Network administration tasks

• Network interface and connectivity: the type of network interface we use, whether access to a local network, a larger network, or broadband type connection with DSL or ISDN technologies. Also, the type of connectivity we will have, in the form of services or requests.

• Data routing: data that will circulate, where from or where to, depending on the available network devices, and the machine's functions within the network; it may be necessary to redirect traffic from/to one or more places.

• Network security: a network, especially one that is open (like Internet) to any external point, is a possible source of attacks and, therefore, can compromise the security of our systems or our users' data. We need to protect ourselves, detect and prevent potential attacks with a clear and efficient security policy.

• Name services: a network has an infinite number of available resources. Name services allow us to name objects (such as machines and services) in order to be able to locate them. With services such as DNS, DHCP, LDAP etc., we will be able to locate services or equipment later...

• NIS (Network Information Service): large organisations need mechanisms to organise and access resources efficiently. Standard UNIX forms, such as user logins controlled by local passwords, are effective when there are few machines and users, but when we have large organisations, with hierarchical structures, users that can access multiple resources in a unified fashion or separately with different permissions... simple UNIX methods are clearly insufficient or impossible. Then we need more efficient systems in order to control all of this structure. Services such as NIS, NIS+, LDAP help us to organise this complexity in an effective manner.

• NFS (Network Fylesystems): often, on network system structures information needs to be shared (such as files themselves) by all or some users. Or simply, because of the physical distribution of users, access to the files is required from any point of the network. Network file systems (such as NFS) offer us transparent access to files, irrespective of our location on the network.

• UNIX remote commands: UNIX has transparent network commands, in the sense that irrespective of the physical connection it is possible to run commands that move information along the network or that allow access to some of the machines' services. These commands tend to have an "r" in front of them, meaning "remote", such as: rcp, rlogin, rsh, rexec etc., which remotely enable the specified functionalities on the network.

• Network applications: applications for connecting to network services, such as telnet (interactive access), FTP (file transmission), in the form of a client application that connects to a service served from another machine. Or that we can serve ourselves with the right server: telnet server, FTP server, web server etc.

• Remote printing: access to remote printing servers, whether directly to remote printers or to other machines that offer their own local printers. Network printing transparently for the user or application.

• E-mail: one of the main services offered by UNIX machines is the e-mail server, which can either store mail or redirect it to other servers, if it is not directed at its system's own users. In the case of the web, a UNIX system similarly offers an ideal web platform with the right web server. UNIX has the biggest market share with regards to e-mail and web servers, and this is one of its main markets, where it has a dominating position. GNU/Linux systems offer open source solutions for e-mail and web, representing one of its main uses.

• X Window: a special model of interconnection is the graphics system of the GNU/Linux systems (and most of UNIX), X Window. This system allows total network transparency and operates under client-server models; it allows an application to be totally unlinked from its visualisation and interaction with it by means of input devices, meaning that these can be located anywhere on the network. For example, we may be executing a specific application on one UNIX machine while on another we may visualise the graphic results on screen and we may enter data using the local keyboard and mouse in a remote manner. Moreover, the client, called client X, is just a software component that can be carried onto other operating systems, making it possible to run applications on one UNIX machine and to visualise them on any other system. So-called X terminals are a special case – they are basically a type of dumb terminal that can only visualise or interact (using a keyboard and mouse) with a remotely run application.