Open Source Network Administration

For a sysadmin, putting "MIT Network Operations" on a resume must feel pretty satisfying. James Kretchmar got the job, and now has written the book. ALecs writes with his review of Kretchmar'sOpen Source Network Administration, below. Open Source Network Administration author James M. Kretchmar pages 220 publisher Prentice Hall (PTR) rating 9 reviewer Joshua Malone ISBN 0130462101 summary A brief tutotrial on using several open source packages to monitor and administer system networks Open Source Network Administration covers a number of open source tools designed to aid in managing computers and TCP/IP networks. The tools discussed in this book are all free, and are all top-quality tools that have earned their place in any system administrator's arsenal of administration and debugging tools. Included in this book are:

• SNMP (a protocol for managing network devices and hosts)
• MRTG (the Multi Router Traffic Grapher - a bandwidth utilization meter)
• Neo (a network device administration tool that speaks SNMP)
• Oak (a syslog watcher and digester)
• Nagios (an active network/host monitoring tool)
• Flow Tools (tools for processing Cisco NetFlow data)

This book also discusses more basic debugging tools such as ping, traceroute, tcpdump and others. Finally, Kretchmar provides some pointers on building your own tools using bash, perl, sed and awk.

Kretchmar is a network engineer for MIT and has gotten a lot of practical experience in managing large networks and unruly hosts. In this book, he imparts a large amount of that experience in over 200 quick-reading, no-nonsense pages. He tells you what a tool can do, how to get it and build it and provides examples of some typical uses. While beginning network administrators will feel comforted that he takes enough time to explain the tools he talks about, experienced ones can safely jump right to his equally well-explained configuration examples without missing anything crucial.

This book read so quickly and was so straightforward that it really inspired me to fix up some areas of my network monitoring that I knew were lacking, but hadn't bothered to fix. In particular, his chapter on Oak motivated me to implement an instant messaging infrastructure (like one he mentions using at MIT) to receive event notices quickly and without dependence on e-mail. While it's no bible (my staple, the Unix System Administration Handbook, is over 800 pages), this book provides a great start on quite a few great tools - many of which I plan to investigate soon.

I was a bit puzzled at his inclusion of instructions for building each tool when most of them are simply ./configure; make; make install. Only one of the tools seemed to actually merit building instructions. At least you can't say he isn't thorough.

I give this book nine stars (out of ten) simply because it really made me realize how easy it is to configure a lot of automation that Ive been wanting. The cover price of U.S. $44.99 strikes me as a bit high in the market, but it is significantly discounted at most online book stores. I still have to recommend The Unix System Administration Handbook first, however. It is more expensive, but contains much more scope and detail than this book. Those who have digested USAH, though, should consider picking this book up from your favorite e-tailer.

You can purchase Open Source Network Administration from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page. Reviewer, Virginia Tech alum and CHUUG member Josh Malone has been a Unix Systems and Network Administrator in Charlottesville, VA for three years.

Comments

Understatement.

[Chris_Stankowitz]

Kretchmar is a network engineer for MIT and has gotten a lot of practical experience in managing large networks and unruly hosts.

I'd say, we all know how MIT loves its pranks/hacks. This has to be one intersting and challenging job. Someone should publish a book (or just a website) on the pranks/hacks that happens on their network. For the most part the other pranks are all well documented, bu I'd love to see what these kids cook up for the New Admin.

Why is EVERY book reviewed as great on slashdot???

[adambehnke]

Honestly, Im not being an idiot. There is no way every book out there is a 9/10, excellent, or whatever. Let's take a look:

Last book review, 9/10

Book before that, rated Excellent

Before that, two thumbs up

Oh, MY GOD, an 8!

What gives? Can we get an unbiased review, please?

Re:Why is EVERY book reviewed as great on slashdot

[prostoalex]

Unlike traditional media, which has a staff responsible for reviewing new books, Slashdot's book reviews are submitted by users. There is little, if any, incentive, to read the books that are worthless or of little value.

Re:Why is EVERY book reviewed as great on slashdot

[jmt9581]

I don't really have a problem with high ratings in book reviews on Slashdot these days, it seems likely that the reviewer really does think that the book deserves 9/10. I think that the lack of low reviews means that people aren't willing to invest resources in buying, reading and reviewing a book that they don't like. I personally will never finish a book that strikes me as low quality. I would also never review a book that I haven't finished.

Judging books before really reading the entire thing is much easier than it sounds. It's very easy to go down to Narnes and Boble and flip through a book and assess it's value within a couple of minutes.

To me, book reviews like this present book reviews to me that I might be interested in. If you really have that much of a problem with the reviews, go find a crappy book yourself and review it. :)

Re:Oak? Where?

[losttoy]

http://perl-oak.sourceforge.net/downloads/Oak-1.8. tar.gz

Link to files

[trboyden]

Here is a link to the files referenced in the book: http://web.mit.edu/ktools/

Regards,
Timothy Boyden
Systems Administrator
MIT Department of Facilities

Re:MIT so great why ?

[kfg]

Charlie Daniels was accused, during an interview, of not being that great a musician and simply hiring people better than himself to back him, making him just look good.

To which Charlie replied, roughly," I'm just a front man. Of course they're better than me. If they weren't I wouldn't have hired them.

Charlie knows something.

You'll never see a job ad that says "Wanted: Really Smart Guy."

No boss really wants that guy around the place. Despite some of your other respondents this fairly common knowledge with lots of anecdotal evidence.

Robert Townsend, author of "Up the Orininization" and former President of both Avis and American Express wrote that every company should have a Bullshit guy. This is a guy whose only job is to wander around and yell "Bullshit!" any time he sees any in a company practice. It' would take a reasonably smart guy to fill this position.

And what boss would have him, wandering around yelling "Bullshit!" at all the policies he implemented?

I take the idea a bit further though. Every company of any size should have a position that's called "The Smart Guy." Officially. That's what it would say on his cards and everything.

No official duties other than being smart, curious, informed and the interest to keep himself informed, on nearly anything.

Anybody in the company could talk to "The Smart Guy." Network's down and you can't figure it out? Go talk to The Smart Guy. Maybe he doesn't know the answer, in fact he probably doesn't, but talking to someone smart might helf you think about the problem in a way that allows its solution.

Extrapolate.

You'll never, ever see this position advertised. Not because it's not a good idea, but because every boss thinks that he's The Smart Guy. He wouldn't be "boss" otherwise. Right?

KFG

Re:What with Longhorn in a couple years

[Random BedHead Ed]

Microsoft has released a lot of crap software, but in this case if Microsoft is working on a product that won't be out for three years, it probably WILL be good, so the comment that the open source community has to keep OSS better than Windows is quite valid.

Consider:

• Microsoft has significantly increased their development staff. By thoustands of developers. My interpretation of this is that they know very well they weren't much good at this before, and that they've been lucky. And now that they actually have to compete with something for a change, they need more people and a better development process.
• The usual baloney about a database-driven filesystem looks like it's actually real this time. WinFS could be a major feature, since most idiot computer users don't comprehend storage. I know - I do desktop support. An intuitive way to find information is much needed by many less-skilled users. This could be a killer app, and OSS developers should probably try to match it, and beat it, prior to 2006.
• Longhorn looks nicer than XP.
• Supposedly M$ has hired former UNIX developers to improve the command line. So the pitiful DOSlike Windows CLI shell might actually be worth using, connecting to remotely, etc.

OSS developers should recognize Longhorn for what it is: Microsoft trying to be competetive for the first time in years. Don't expect another crap OS from them.