Martin Ansdell-Smith

mas system administration links

DevOps, DevSecOps, SRE

Back to top of page

Distributed systems

Back to top of page

Distributions of GNU/Linux software

Back to top of page

Magazines and podcasts

Back to top of page

Monitoring individual systems

GNU/Linux systems have a rich variety of monitoring tools. Ones I use regularly include:

In addition, for VPSs, the vendor may provide statistics, graphs and reporting tools. These may use tools such as those above to provide data to their customers.

16 basic server monitoring commands by Steven Vaughan-Nichols. A reminder of basic tools you should be using anyway. Includes iostat, meminfo (cat /proc/meminfo), free, mpstat, netstat, nmon, pstat, ps, pstree, sar, strace, tcpdump, top, uptime, vmstat, wireshark.

Back to top of page

Security

Back to top of page

Services

User groups and trade associations

Back to top of page

Systems administration

Systems administration is a fundamental art and science for human interaction with general purpose computers. At its widest it includes everything to do with computers that isn't actually writing and using the application software. For the biggest systems this may be termed infrastructure management with systems administration being working on individual machines but this boundary is somewhat artificial and has to be flexible. A computer, whether an individual machine, multiple machines, or a shared multiprogram and multi-user system, must be capable of doing the jobs needing to be done but it also needs to be suitably affordable (both to buy and to run), environmentally sound (power, heat, size, weight, noise, raw material sourcing and disposal), secure, resilient, robust, reliable, speedy, etc. As always one probably cannot have all the desired qualities at the same time … today. Some desired, or even necessary, attributes may be impossible to fully meet.

For individual machines, including virtual ones, system administration includes requirements analysis and specification, initial configuration and set-up. It continues throughout the life of the system with ongoing updates, fixes, backups, monitoring and enhancements. It ends with the shutdown of the system, possibly archiving its contents, termination or transfer of any leases, licenses or rentals. It initiates disposal of any redundant hardware.

Systems administration includes or requires related skills and disciplines such as systems programming, security administration, facilities management (e.g. power, air-con, accommodation, access control) and capacity planning. Networked systems also require network design, network engineering, network administration and network analysis.

For multi-machine systems the needs and requirements are similar but achieving good outcomes becomes more complex and difficult.

The modern trend towards cloud services has changed the nature of some of these steps but they still need to be done. If you are not doing them for yourself then you are trusting someone else to do them well on your behalf: a cloud service is just your data entrusted to someone else's computer.

Back to top of page