Most Helpful Customer Reviews
22 of 24 people found the following review helpful:
5.0 out of 5 stars
Wow, June 13, 2005
This review is from: Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software (Paperback)
You ever find a reference that becomes your best friend - that you can't put down? I looked through three other books on clusters, including Oreilly's... They were all good, but missing the "rubber meets the road" factor.
I have the proverbial "Wall of Oreilly" in my office. But this book wins out for wide clustering.
I needed to deploy a highly-available, fault-tolerant, hi-performance cluster of cheap servers so that I could sleep at night knowing that my web services were live - and would remain that way.
This book met the challenge head-on. Mon, cron, init, Linux Virtual Servers (IPVS/LVS), SNMP, Heartbeat, etc are all covered with a "here's how you do it" approach.
My only problem with the book was that the examples were all for Redhat, which I honestly don't agree stands up to the true intent of Open Source anymore... I think the book should be directed at Debian instead.
Anyway, the book was still great for me as most of the packages used are similar between platforms.
There ARE typos. In fact, I found a few critical ones in code snippets and some confusing ones in the descriptions where it looks like the paragraphs got mixed up, so be sure to look at the snippets with more than a cut&paste attitude. (if the author is looking for edits for another release/update, please contact me - I happen to be a copy editor and a programmer)
IMHO, THIS is the future of linux server computing. Please support more books like this by buying it. Honestly, I might actually look into other books by "No Starch" based on my good experience here.
BTW, this book is devoted to what I consider "WIDE" clustering (LVS/IPVS - processing allot of small computations - focus is on response time) versus other books focused on "DEEP" clustering (Beowulf - processing fewer but more intense computations - focus is on CPU throughput). So if you're looking for the latter, look elsewhere.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
21 of 23 people found the following review helpful:
5.0 out of 5 stars
One stop guide to building a Linux Cluster, July 18, 2005
This review is from: Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software (Paperback)
The sheer number of books published in the recent years that cover the Linux Operation System is simply staggering. One aspect of Linux that has been explored recently is its enablement that is gives users that wish to build a clustered environment. Using free software and commodity hardware many organizations small to large, and even end-users such as myself, have been able to build a cluster or a Grid using Linux very cheaply. This book by Karl Koppers shows the readers a step-by-step guide of how to go about building an enterprise grade Linux cluster environment. The author depicts each step one by one for the reader and demonstrates what each step does and why "we do what do we".
The following steps required for building a cluster are explored in detail in this book:
* Understanding of the underlying Linux routing (NetFilter and basic packet routing)
* Understanding of SystemImager program, and how it can help you in building a large cluster
* Learning of to apply configuration changes to your cluster efficiently
* Learn to build a Linux Virtual Server Network Address Translation fro your cluster, and learn why you need such capability
* Learn how to manage failures in your cluster including software and hardware failures
* Learn how to manage and monitor your cluster from a single point using a Web-based GUI
* Learn how to apply software patches and updates to your cluster
* Administrator user accounts and remote login including password-less login via SSH
* Learn how to install a cluster print server
* Learn how to install a batch-job scheduling system on your cluster
As you can see, there are lots covered in this book. Some of the basics of the underlying Linux system and Kernel gets the reader off on a good start, and starts the foundation upon which the rest of your cluster can be built.
Unlike any other book that claims to cover this topic, Karl has added a number of chapters on the understanding (he calls them theory of ...) of the concept and why things are done the way they are instead of just telling the reader what to do. These chapters can be skipped, obviously, but I found them to be extremely useful because it gives me a deeper knowledge of the topic at hand makes me understand the topic better. The whole point of the book is to enable someone, by simply going thru the book chapter by chapter, to build a medium to a large size cluster. A how-to guide that is packed with the appropriate software packages discussed in the text.
My favorite topics in this book must the central administration of the cluster via Ganglia, and High-Availability configuration chapters. These two, monitoring and reliability, are key difficult areas which the author covers in detail.
The combination of the book and the accompanying CD-ROM packed with software packages make this book a great buy. I have built my own cluster, but I am re-doing my environment by going thru this book chapter-by-chapter. I highly recommend this book to anyone involved or interested in building or maintaining Linux Clusters.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
10 of 10 people found the following review helpful:
4.0 out of 5 stars
cheap clusters, May 21, 2005
This review is from: Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software (Paperback)
The book offers practical advice in putting together the hardware of several linux machines. Such that you end up with one big and robust computer. The attraction is price. The machines use commodity hardware and linux is free. As opposed to an expensive monolithic computer.
The book's audience appears to be a sysadmin or network designer. Very details oriented. Much attention is given to system utilities that you will need to monitor this network that is the cluster. A big benefit of the book is in simply educating you as to what utilities have already been written under linux, to support clusters.
The clustering described in the book does not seem to encompass mentioning Beowulf clusters. The latter is an alternative approach that has seen some adoption in scientific circles. I'm guessing that from what the book describes of a linux enterprise cluster, that the omission might be because linux now has the equivalent capability.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
|