Most have forgotten that the Internet Engineering Task Force originally recommend OSI with full implementation of all 7 layers of the ISO model. Of course, no one wanted that...
You got what you asked for !
I worked for a Fortune 100 engineering/manufacturing company as a Sys Admin for over 20 years. In my last position I worked in a small niche of the company outside of major machines and applications but still in a place where a system problem that lasted more than 24 hours could shutdown multiple factories. My immediate management did not want to know anything. My favorite quote was, "The best you can do, is to not let my phone ring with a complaint !" Any management above that level (incorrectly) assumed that the PCs on engineers desks were all that was required for them to get their work done when in fact all of their work was conducted on servers under my control.
Worse, the servers in use averaged between 8-12 years old, some running "retired" OS's, most of which had no hardware maintenance contract as we had "spares" or could reconfigure them with less CPUs/memory until replacements could be obtained from the used market.
trims did a great summary of different jobs in the Unix field. After 26 years in the computer biz (hence the nickname) I'll throw in some comments.
Different employers want different (combinations) of talents. Big companies (like where I work) have things very "compartmentalized". I'm not allowed to touch any network gear and (thank God) I don't have to do any WinDoz support. I do some VMS support, but that is where I cut my teeth.
We have a couple of app programmers, a Webmaster, a SCM guru and a QA "jack-of-all-trades" guy. My "sidekick" and I admin about 70 boxes, running 2 different flavors of Unix, in 3 different buildings, in 2 different cities (10 miles apart). We are responsible for installing/tuning/cutomizing the OS and GUI (more later). We also "build" gnu/public domain apps like perl and some of its infinite modules, a2ps, lsof, emacs, nedit, etc, etc. We also install all commercial apps.
Most small companies want you to "wear multiple hats", or maybe even all the hats that trim mentioned ! Breadth of knowledge is frequently more important than depth. Unfortunately, the School of Hard Knocks is frequently the best instructor.
Unix never "lives" alone these days. Either you have WinDoz or IBM Maninframes or VMS or some "odd" flavor (QNX, AIX) that you have to co-exist with. You'll have to learn enough about the other guys jobs to talk to them. The same applies with networking.
From a sys admin perspective the most difficult thing I had to "grasp" was automount. KISS applies here, especially if you have an even moderate number of systems. At my site we basically have one "big" machine NFS serving all the rest. Easy to understand and back up.
Anyone still using fsck on Solaris (8 and up) needs to do a man mount_ufs and check the logging option. I'm not knowledable of Linux, but I think any "modern" Unix should have a log based file system (which means you never have to fsck again).
Working in "different" Unix environments at the same time is a pain. Solaris will tell you that System V is the only way and then hide all of their XPG4 compliant stuff in other directories. Carefull PATHing kept my user base (some who still can't spell Unix) from string me up. I hate the SVID printing so much (how many times has Sun re-written it) that we have built our own BSD printing subsystem on Solaris (MUCH easier to admin).
Lastly, I would like to kill the folks who invented CDE. It is a pain to admin/configure on a large scale. Give me a simple "laucher" and a Motif window manager and I'm happy.
Very last. Know enough hardware that you can talk to your support folks.
Enough or I'll start talking about 36 bit computing !!!
Slashdot Mirror
Most have forgotten that the Internet Engineering Task Force originally recommend OSI with full implementation of all 7 layers of the ISO model. Of course, no one wanted that ...
You got what you asked for !
I worked for a Fortune 100 engineering/manufacturing company as a Sys Admin for over 20 years. In my last position I worked in a small niche of the company outside of major machines and applications but still in a place where a system problem that lasted more than 24 hours could shutdown multiple factories. My immediate management did not want to know anything. My favorite quote was, "The best you can do, is to not let my phone ring with a complaint !" Any management above that level (incorrectly) assumed that the PCs on engineers desks were all that was required for them to get their work done when in fact all of their work was conducted on servers under my control. Worse, the servers in use averaged between 8-12 years old, some running "retired" OS's, most of which had no hardware maintenance contract as we had "spares" or could reconfigure them with less CPUs/memory until replacements could be obtained from the used market.
trims did a great summary of different jobs in the Unix field. After 26 years in the computer biz (hence the nickname) I'll throw in some comments. Different employers want different (combinations) of talents. Big companies (like where I work) have things very "compartmentalized". I'm not allowed to touch any network gear and (thank God) I don't have to do any WinDoz support. I do some VMS support, but that is where I cut my teeth. We have a couple of app programmers, a Webmaster, a SCM guru and a QA "jack-of-all-trades" guy. My "sidekick" and I admin about 70 boxes, running 2 different flavors of Unix, in 3 different buildings, in 2 different cities (10 miles apart). We are responsible for installing/tuning/cutomizing the OS and GUI (more later). We also "build" gnu/public domain apps like perl and some of its infinite modules, a2ps, lsof, emacs, nedit, etc, etc. We also install all commercial apps. Most small companies want you to "wear multiple hats", or maybe even all the hats that trim mentioned ! Breadth of knowledge is frequently more important than depth. Unfortunately, the School of Hard Knocks is frequently the best instructor. Unix never "lives" alone these days. Either you have WinDoz or IBM Maninframes or VMS or some "odd" flavor (QNX, AIX) that you have to co-exist with. You'll have to learn enough about the other guys jobs to talk to them. The same applies with networking. From a sys admin perspective the most difficult thing I had to "grasp" was automount. KISS applies here, especially if you have an even moderate number of systems. At my site we basically have one "big" machine NFS serving all the rest. Easy to understand and back up. Anyone still using fsck on Solaris (8 and up) needs to do a man mount_ufs and check the logging option. I'm not knowledable of Linux, but I think any "modern" Unix should have a log based file system (which means you never have to fsck again). Working in "different" Unix environments at the same time is a pain. Solaris will tell you that System V is the only way and then hide all of their XPG4 compliant stuff in other directories. Carefull PATHing kept my user base (some who still can't spell Unix) from string me up. I hate the SVID printing so much (how many times has Sun re-written it) that we have built our own BSD printing subsystem on Solaris (MUCH easier to admin). Lastly, I would like to kill the folks who invented CDE. It is a pain to admin/configure on a large scale. Give me a simple "laucher" and a Motif window manager and I'm happy. Very last. Know enough hardware that you can talk to your support folks. Enough or I'll start talking about 36 bit computing !!!