Customer Reviews

The Art of Capacity Planning: Scaling Web Resources

5 star:		(7)
4 star:		(2)
3 star:		(3)
2 star:		(0)
1 star:		(0)

 

 

John Allspaw has done something that very few of his peers would have been able to do. He has taken a black art, Capacity Planning, and he turned it in to a series of steps that anyone can follow.

The book is filled with common case studies, for how to plan capacity for things like web server farms, database clusters, and caching layers. The real value is in watching how the author applies the same formula in each case, giving Systems Administrators and Executives the tools they need to do a better job of capacity planning in their own unique infrastructures.

As the earlier review says, it's a short book. In my opinion, that's a good thing: it's goal is to teach you how to perform capacity planning in any environment. If it was longer, it would have been full of more examples, which would likely only serve to lead the reader away from the core principles. You need to learn *how* to capacity plan an infrastructure, not get pat (and often incorrect) advice on how to measure your web farm.

The discussion on curve fitting and trend prediction is worth it alone - I'm aware of no other book on the topic that shows so clearly how to examine your data in service of capacity planning.

It's the process I'll follow from now on.

The Art of Capacity Planning is a good introduction to Capacity Planning for Web Operations that touches on the following topics:
* Why do you need capacity planning?
* What information should you gather for capacity planning and how?
* How to predict trends for your web applications?
* How and when to procure new hardware?
* How to create a sustainable capacity planning process?

As the author mentions in the preface, the book has a lot of common sense material. Most experienced enterprise web operations architects should be familiar with this material. But, it is refreshing to see this urban wisdom captured and printed in a book format. The book is unique in that it is not meticulously organized and illustrated like a text book or a reference guide. It provides a smattering of anecdotes, examples, gotchas, and tools from the author's experience in a rapidly growing start up environment at Flickr.

I am looking forward to a second edition of the book where the author can delve deeper into some missing aspects that are critical to capacity planning like log analysis and performance improvements. Enterprise web operations folks who are familiar with commercial tools like Sitescope, OpenView, Opsware, Gomez, etc. rather than free/open source tools and who manage a large number of diverse applications might have a learning curve to relate the examples in the book to their environment.

Right out of the gate, John covers a topic near and dear to my heart: metrics. His advice? "Measure, measure, measure." John reinforces this by including an incredible number of charts throughout the book. He goes on to say that our measurement tools need to provide an easy way to:
* Record and store data over time
* Build custom metrics
* Compare metrics from various sources
* Import and export metrics

As I read the book, I found myself nodding and thinking, "yes, yes, this is exactly what I learned!" Although it's been more than five years since I was buildmaster for My Yahoo!, I really resonated with the advice John provides, like this one: "Homogenize hardware to halt headaches". (You have to love the alliteration, too.)

In a thin book that's easy to read, John covers a large number of topics. He talks about load testing, with pointers to tools like Httperf and Siege. There are several sections that talk about caching architectures and the use of Squid. He provides guidelines when it comes to deployment, such as making all changes happen in one place, the importance of defining roles and services, and ensuring new servers start working automatically. At the end he even manages to cover virtualization and cloud computing, and how they come into play during capacity planning.

The Art of Capacity Planning is full of sage advice from a seasoned veteran, like this one: "The moral of this little story? When faced with the question of capacity, try to ignore those urges to make existing gear faster, and focus instead on the topic at hand: finding out what you need, and when." When I read a technical book, I'm really looking for takeaways. That's why I loved The Art of Capacity Planning, and I think you will, too.

This is the first book on capacity planning I have read so I have nothing substantial to compare it too at this time. John's descriptions and real world examples are great.

While I was reading I felt John's analogies were very similar to the way the character Charlie from TV's "Numb3rs" explains something very complicated with a real world examples. I liked the examples of the Bacon Delivery truck and the Super-market checkout especially to visualize what was going on in the process of the servers.

One huge take away was the level of importance tying application metrics and server metrics back to financial costs. SLA's don't really matter if the cost of adding another 9 to the 99.999's type model is more expensive than your client is paying you for the whole contract. In essence don't promise 99.9% over 99.0 percent if the .9 improvement will cost $10,000 in additional hardware and the contract is only worth $10,000. Many would argue but it is only a 9/10ths of a percent improvement how big of a deal can it be? Remember the first 1% of keeping up a server is not the same as the last 1%.

The chapter on regression and line fitting was mostly a refresher. The chapters on cloud computing were excellent as real world examples are always useful for me. I also liked the fact he referred to flickr a lot, so there was a sense of walking the path vs. knowing the path.

Some co-workers did joke that they must not know what they are doing because the seats are all empty on the cover. I'd be curious to see if the same book sold better with the same cover and seats filled. Other comments criticize the book for being only 150 pages but I would rather have 150 good pages than 300 bad pages any day of the week. Also the author explains the smallish size in the preface.

All in all a great quick read that cut to the details and made me feel more confident I could bridge the gap between business and IT in a short amount of time.

The "Art" is an approachable treatment of a complex field of operations: capacity planning for high-traffic websites. Allspaw leverages his Flickr experience to give us a window into web operations as done by the pros.

The book keeps the high-level perspective necessary to give useful advice in a messy field, without getting lost in minutiae that would be specific to a given site. The author goes over the hows and whys of planning your capacity and the process needed to maintain it as traffic grows, with interesting insights such as designing for measurement (i.e. not mixing separate components of the architecture on the same machine in ways that hinders measurement of actual capacity), how to place a procurement process in place, and the ever-present point of presenting your data convincingly to the business owners that write the checks.

Allspaw places the emphasis on the right places, and does so in a concise manner: at less than 150 pages, this book packs a lot of meat for its pages, and as a fan of brevity the point did not go unnoticed on me. This is one of the best titles to come out of O'Reilly in the last few months, a must-have for your technical library if you work in the field.

Capacity planning is one of those areas where not many books are published yet.

One reason might be that it's a complex task. You need the right mixture between a certain practical approach, measuring the right system parameters, putting them into meaningful context, making educated guesses and doing the right maths.

The author(s) shows in his book examples on how to gracefully plan extending your server farm. Examples show how to find the limites of your web, cache, DB, storage or image processing server and how to mitigate or live with them.

As you can see the book is trying to cover a lot of ground.
Its also going into discussions on curve fitting and trend prediction.
Its even going into a description of cloud computing in the appendix.
Overall the author(s) trys to give many useful advice on a sitenote which actually deserve more detailed exploration.

The fact of the books "diet" from the originally announced 300 to only 150 pages doesn't work for its advantage. A few more case study scenarios or more detailed instruction like tutorials would have proven useful.

So its mostly up to the reader to exercise and follow up on the author(s) intentions.
That's also what will make the mileage vary for many readers.

Unfortunately, I got the impression that the book is rushed to the marcet.
F.e. the search index contains no page numbers.

For the PDF version I thought the Images are at times difficult to read.
When opening the PDF on Linux you might find the images are not displayed correctly.

The book serves as an extension to the public session "Capacity Planning" which you might have seen on the Velocity Event on 23rd/24th June in Burlingame.
h**p://en.oreilly.com/velocity2008/public/schedule/detail/3208

The compact book is split into two major parts.

The Main Body - 110 pages
The Appendix - 22 pages


Summary:

The overall idea of this compact book is sound.
It covers topics that hasn't been explained anywhere before.
The wide coverage of the book is a blessing and a curse at the same time.

On the one side there are a real gems hidden in the text, especially in the 20+ page appendix which is even going into cloud computing.
On the other side the author(s) do give many useful advice on a sitenote which actually deserve more detailed exploration.
That's why, the book needs to major to a second edition to really bring out the author(s) original intention in full.

As of now (1st edition) the book does serve as a lightweight overview of CP for consultants and managers. From a system engineers point of view I feel the book should have delivered more on how to perform capacity planning.

John Allspaw's book is a classical treatise about capacity planning. It is a valuable asset to all web services professionals. John's lucid style of coupling statistical concepts to the capacity planning (and prediction) practice makes the book very special. The book is methodically organized, in that it takes the reader gradually from a generic view of a capacity planning project to the specifics of the measurement, monitoring and forecasting of web traffic. The book also has a valuable source of information about various statistical tools that help in analysis, design and upgrading of the capacity of web sites. The author's awesome presentation (using mathematical statistics) of all the steps involved (a) in the measurement of current traffic metrics and (b) in forecasting the capacity of a web site, deserves a special word of praise. In summary, I would recommend this book to all professionals involved in the building and maintenance of the web infrastructure, for a business.

This book is fantastic! It's a very easy read, keeps you engaged and is full of excellent advise for all stages. Architectural planning, monitoring, trending, and utilizing public clouds for your occasional peak cycles.

I can't recommend this enough, my team at work is now reading this and our conversations have changed for the better.

The Art of Capacity Planning is a good introduction to Capacity Planning for Web Operations that touches on the following topics:
* Why do you need capacity planning?
* What information should you gather for capacity planning and how?
* How to predict trends for your web applications?
* How and when to procure new hardware?
* How to create a sustainable capacity planning process?

As the author mentions in the preface, the book has a lot of common sense material. Most experienced enterprise web operations architects should be familiar with this material. But, it is refreshing to see this urban wisdom captured and printed in a book format. The book is unique in that it is not meticulously organized and illustrated like a text book or a reference guide. It provides a smattering of anecdotes, examples, gotchas, and tools from the author's experience in a rapidly growing start up environment at Flickr.

I am looking forward to a second edition of the book where the author can delve deeper into some missing aspects that are critical to capacity planning like log analysis and performance improvements. Enterprise web operations folks who are familiar with commercial tools like Sitescope, OpenView, Opsware, Gomez, etc. rather than free/open source tools and who manage a large number of diverse applications might have a learning curve to relate the examples in the book to their environment.

Great content. The size of the print is so small it distracts from reading the text. [...] you should be able to read the damn book easily. Is O'Reilly conducting a Dilbert-ish shrinking page count experiment?