[Ericsson AB]

1 Introduction

1.1 Purpose

In the most perfect of all worlds this document would not be needed. The compiler would be able to make all necessary optimizations. Alas the world is not perfect!

All considerations for efficiency are more or less implementation dependent. Efficient code is not always good code from the perspective of generality, ease of understanding and maintaining. Therefor programming becomes a balance act between generality and efficiency. So how do we manage to walk on the lim and not fall off? Well, on a structural level there are things that you can keep in mind while you design your code. This guide will try to help you use data structures and mechanisms of Erlang/OTP in the intended way, this will help you avoid many unnecessary bottlenecks. Apart from those structal considerations, you should never optimize before you profiled your code and found the bottlenecks. Also remember not all code is time critical, if it does not matter if takes a few seconds more or less there is no point in trying to optimize it. Profiling erlang code is easy using tools such as eprof, fprof (from release 8 and forward) and cover. Using these tools are just a matter of calling a few functions in the respective library modules. Taking the appropriate measures to speed the code up once you found the bottlenecks can be a bit harder. You may have to invent new algorithms or in other ways restructure your program. This guide will give you some background knowledge to be able to come up with solutions.

Note!

For the sake of readability, the example code has been kept as simple as possible. It does not include functionality such as error handling, which might be vital in a real-life system. Inspiration for the examples is taken from code that has existed in real projects.

1.2 Pre-requisites

It is assumed that the reader is familiar with the Erlang programming language and concepts of OTP.


Copyright © 1991-2006 Ericsson AB