Rating:

The book is a nice introduction to fault-tolerant software design. I liked the concepts introduced and will think about some of them when applicable.

For example, the book suggests using a pattern called ‘fault-correlation,’ which tries to determine the correct mitigation strategy based on several factors that make up the signature of a fault. Given that signature, one can ‘categorize’ a fault and apply pre-defined mitigation steps. I’ve never used a technique like this, and I am unsure whether I could have come up with it myself.

I did not like the pattern style of the book very much. There are so many patterns, and I felt that not enough consideration was given to each.

• The number of examples per pattern got fewer as the book progressed<
• Some patterns make almost no sense on their own and are tightly bound to others. E.g., there is a complete pattern called ‘Reassess Overload Decision,’ which just advises reassessing a decision made in another pattern. Is that really a pattern on its own?

I didn’t read the book until the end, so maybe it gets better in the last parts.