Second edition of TPOSANA

[yesthattom]

Last week we signed a contract to write a second edition of The Practice of System and Network Administration. It isn’t going to be a big update, just a lot of much-needed updating. We’re also adding a new co-author. I don’t think we’ve announced who it is yet, so I’ll keep you all in suspense.

But what I really want to tell you in this LJ entry is:

This morning I sat down and started reading it from page one. I took a spare copy and started marking it up. Oh my god, the first 30 or so pages suck. How was this book a success when all the stuff at the front sucks so hard? How did anyone ever get to Chapter 2 when it starts to get better?

And the fact that the book is so Solaris-centric is bizarre. The world has really moved to Linux and I don’t think we mention Linux except rarely.

P.S. If you have suggestions, corrections, or want to point out things we missed the first time around, etc. let me know!

Comments

[figmo.livejournal.com]

I know who the 3rd author is, and IMHO the 2nd edition is going to rock. (I don't know whether it was announced; I have my sources.)

[kimuchi.livejournal.com]

I suspect it looks a lot worse to you than to anyone else. But also I know I've used it in a disjointed way rather than from cover-to-cover (aside from when I was trying to do pre-release reviewing), and I suspect that's common with technical books as a whole.

Solaris really was The Thing when you guys wrote this the first time, remember. I don't think I'd touched a single linux machine in a professional capacity by 2000 (I had worked with FreeBSD, BSDI, and SCO a bit, though).

[rainbear.livejournal.com]

Well, given the fact that we were using FreeBSD, and you had come from a Solaris background..and Solaris x86 was coming out.... I think it was reasonable that you hadn't really touched on Linux. Plus, seriously, Linux has matured a *lot* since then.

I'm looking forward to the new edition :-) (And I think I know who the new co-author is, but hey, I have my sources too (grin)!)

[cpj.livejournal.com]

I'll have to re-read those chapters to try and get your perspective on this.

What exactly is Solaris specific? Command invocations? Architecture issues? I never really noticed a bias before.

When it comes to using either OS, the administration principles are pretty much identical; there are just a few nuances in specifics. What usually applies to one OS will also apply to another.

I think you're going to see a resurgence in Solaris on the x86 side, if you haven't seen it already. They have better tools, better development practices, and a better focus on where to take the OS. The big downside is there is still a lack in third party driver development.

[rainbear.livejournal.com]

...and couple that with the fact that the book was more about "practices" as opposed to "This command does this.. that command does that..."--either way, I'm looking forward to the new edition :)

[yesthattom.livejournal.com]

All the examples are Solaris this and Solaris that. I even give the example of distributing an updated Solaris /etc/ntp.conf file.

It isn't that things would break if tried on Linux, it's just that I could have said "Unix" instead.

[petdance.livejournal.com]

Let me know when you want my comments. I've marked mine up in places.

[yesthattom.livejournal.com]

Email them any time. Sooner is better.

[cpj.livejournal.com]

Oh yeah. Do you want some datacenter pics??

[hcoyote.livejournal.com]

I look forward to the update.

One of the things that we've been looking around for is recommendations for running and implementing high density data centers (eg, greater than 300 watts a square foot). I'm not sure if it's a good fit for the book, but heck, it's a suggestion. :-)

[sacristan.livejournal.com]

I've got some useful Layer 0 (a.k.a. "Nice router. Pity we've nowhere to put it") notes from a SAGE-IE talk I must update. Had to plan a 8kW/rack build since then, which was fun...the cluster vendor initially specified 16kW/rack, which was frickin' terrifying...

What do you mean, we can't add layers to the ISO model? We already added Layer 8 (budget), layer 9 (politics) and layer 10 (who the CEO plays golf with).

[hcoyote.livejournal.com]

Yeah. We're already doing 11kW/ rack ... which puts us around 150 watts/sq ft. I know Intel is doing atleast 525 watts/sq ft with forced air. It's a highly tuned environment. I think they were up around 24-27 kW/ rack. I'd have to check the notes from the high density computing symposium that a coworker went to.