Building Linux Clusters

Author: David H. M. Spector



Publisher: O'Reilly

Reviewed by: Dean Wilson

Building Linux Clusters was a book that I had high hopes for. Clusters are one of my hobbies and when I discovered that the same publisher that bought Running Linux to me was behind I saw good things ahead. And then I got it.

This book was held back for months from its initial release date. I can understand this since pretty much anything in the Linux word is a moving target and around the release date clusters were a prime example so when I got my copy of this book I was looking forward to reading about a subject I have an interest in. I was even feeling forgiving if some of the details seemed a little dated.

Dated details wasn't the problem, a book that I had trouble not throwing in the bin was. The whole book smells of padding, and bad padding at that. The author leaves some of the aspects unclear (P83 he mentions the at command and in a comparison to the batch command adds "presumably the more CPU resources the job may use." Now I will admit that I've not put it in very good context but I dislike the word 'presumably' in that sentence. Consider that this man has been paid to release a technical book and he should have either left the statement out or checked it. Technical error no, sloppy and off putting yes.

Another of my favourites is the fact that the book spends over half a dozen pages detailing how and where to set the cluster up, points such as "make sure the cables are away from sharp edges to ensure that they don't get cut" are mentioned. I'm wondering if the editor removed the section that said not to run with scissors while building the cluster. While I can understand and agree with the need to mention considerations such as air conditioning and reinforced flooring the assumptions that are made about the audience are inconsistent and end up alienating the more experienced users while confusing the readers who have picked this up as a first tome.

Take this winner, "If you are mixing technologies (such as 10 MB and 100MB Ethernet) there will be a large difference in speed" Well duh. I'd hope that anyone who was looking at deploying a cluster of a size that would need reinforced concrete flooring would be aware of this. While a book needs to be open to different skill levels I feel that in trying to make this a good starting place for one and all the author has in my view failed in this and made the book an overly difficult read. A subject like clusters is IMHO a moderately advanced one and the author should be able to make some assumptions about the skill level of his target audience and not end up having to mention that 10 and 100 BaseT are different speeds.

I'm annoyed that at least the first half of this book is a complete waste of time and the fact that I paid over 25 pounds for it.

While re-reading this text I've noticed how harsh this review comes across and that I sound like an arsehole. But to be honest this is one of the few books I wish I'd never bought and I just hope that the second edition (If one ever appears) has a lot more work put into it. Then again I'm not a published author and so these are only my views and what do I know.

Summary: A wasted opportunity and a let down from a normally great publisher.

All views are my own and if you disagree write your own review :)

PS At one point the book has included pictures of plug labels... I'm not making this up!!!