Customer Reviews

Scalable Internet Architectures

5 star:		(9)
4 star:		(5)
3 star:		(1)
2 star:		(1)
1 star:		(2)

 

 

I picked up this book the other day on the recommendation of a friend, and I can't put it down. I like it because:

* its small (the number of pages 225, the print, the format, the thickness) , but its pithy. Every page has useful stuff.
* the real world experience (pain!) just oozes out of this book. So many times while reading I thought: "Oh yes ... hadn't thought of that."
* its not stridently opensource, nonetheless ends up most there anyway - but only after addressing commercial products and their role
* its business focused, not geek focused - while still being incredibly geeky

Most of all its just really well written and edited. I've caught a couple of minor typos, but nothing that interfered with reading or enjoying the book.

In a world of many great technical books, and insufficient time to read them, its hard to know which ones are worth the effort. This one definitely is.

This is a fantastic book about scalable systems. If you want specifics, he presents opensource and cost effective solutions that can be implemented, but in my opinion that isn't the real value of the book.

The real value is the way he leads you through the thought processess that need to occur as you plan for releasing and using such a system. I really like some of the stuff that is emphasized and has caused me to realize that I had gaps in my knowledge. Gaps like better release planning, and actual cost of such a system, especially as it grows, or shrinks.

I've been extremely happy with this purchase.

This is the type of technical book that comes alone all to infrequently. Instead of giving "cookbook" recipes that are inapplicable to the majority of real-world environment, this book discusses how to apply techniques (not recipes) to increase the flexibility (and hence scalability) of your infrastructure. This author obviously has in-depth knowledge of real-work production environments and the unnecessary risks that companies expose their infrastructures to. For example, from my own experience it is pitiful how rampant the concept of HA is confused with load balancing! I have seen CIOs of Fortune 500 companies tout their "risk-free" scalable environment implemented with load balancing. When I explain to them that HA has nothing to do with load balancing, and to insure a high availability infrastructure they need to revamp their architecture, they look at me suspiciously as if I'm trying to squeeze money out of them. When I ask to speak to the Director of Operations and pointedly ask how HA is implemented, and get a response alone the lines of "we have redundant load balancers fronting redundant servers" I know I'm dealing with yet another instance of a gross misunderstanding of HA. When I point out the multiple single points of failure, I'm treated as an adversary intent on discrediting the technical staff! This book reveals the fallacy of such an approach.

The author also discusses elements of a robust HA environment that others book only touch on if not completely ignore. For example, revision control and business continuity among other often overlooked processes. All in all an interesting read that will open your eyes to what is truly a misunderstood topic.

The books handling of scalability, including a discussion of "scaling-down" (or scaling back), is first rate as well.

First and foremost, chapter 10 should be an Appendix. This was a horrible ending to what seemed to be a promising discussion on horizontal scaling for any system/network engineer or astute systems engineer.

Clear and concise, then incoherent and grammatically challenged, this book requires constant read backs leaving the reader with a sense of a diminished level of reading comprehension.

Fortunately there are some very good real world discussions on horizontal scaling, distributed caching, and eliminating single points of failure in your design.

Unfortunately the book was a long documentary on the author's Spread utility/program/solution and the last chapter is dedicated to writing a module for Spread. Completely out of band with regards to actual high performance clustered environments where the author's solution is likely scarce in popularity.

I do appreciate his coverage of logging. Despite my rating, I don't regret the first nine chapters.

For those companies wanting to make available their assets on the Web, on a global scale, then this book can help. It describes the many ways that you can perform this scaling.

Like setting up image clusters that are geographically distributed. And applying methods that deliver packets to an IP address based on the shortest distance to one of a set of servers. A particular instantiation is called Anycast. But the idea is generally applicable.

Another issue is whether to have static or dynamic caching. Depending on the circumstances, one might be significantly cheaper than the other.

The book deals well with these and other issues.

This book does a great job of demystifying many of the misconceptions surrounding web applications, performance, scalability, hardware, software and proper architecture. Finding the right tool for the job is essential, and this book should be on every developer/engineer/managers book shelf!

An excellent book on what the issues are with building web site that scale to large numbers of servers. Written more for a programmer than a systems administrator but not to heavy on the programming side to bog it down. It is really good about organizing the different aspects of scalability and explaining important concepts. For example the difference between caching static content and caching data used in generating a web page. An excellent read. not too long, not too much detail. It made me think about web sites and clustering in a new way.

I bought this book on recommendations from others and I have to agree that it is fantastic. Don't let the 2006 publication date fool you into avoiding it for something more current. The advise given is based on real experience gained initially during the dot com era but it is as relevant today as then. The best part of the book is its advocacy and practical examples of the Spread group communications toolkit. Why solve a problem with a vendor's expensive high performance single point of failure solution when well knowledgeable use of internet infrastructure and peer to peer communication better solves it. Plus, you have a powerful tool to aid you moving further towards automatic, self-managed operations.

I was disappointed in this book. Even if it does seem that it covers all the right things, in right order, trying not to miss anything -- down the road it came up clearly that this book is too wordy, too much pouring the water, discussing and explaining same things all over again. It did seem for me that it's a book mostly for system or network administrator, but it pretty useless for software architect. So, for me, the title of the book was misleading. I cannot say that it is useless for everybody, but it certainly was useless for me, and absolutely not the book I was expecting to see. Nevertheless, for the sake of truth I would like to say again that it does contain some interesting material.

This book is great, especially as a starting point to learning about scalability planning. It takes a fairly academic approach to all relevant subjects and can quickly get anyone caught up to speed.

I do recommend "The Art of Capacity Planning: Scaling Web Resources" as a follow-up for continuing your education in this subject and taking these skills to a more proactive plan/deploy method.