Due to its SQL-based engine, the OpenSource Software pgRouting is the most flexible routing engine available. A common misconception is that pgRouting is the least performant routing engine. So how to keep both performance and flexibility with pgRouting ? Many factors should be taken into account. To begin with, the use cases. What level of precision do you need ? Will you be computing short or long routes ? Will there be many routes computed at the same time ? Simplification is the most common way to deal with performance issues. However, when accuracy is at the core of decision making, a minimum level of precision must be kept. Reducing the number of rows the routing engine will have to process is the number one tip to enhance performances. But there are many other technical and functional optimizations that can make pgRouting run much faster. We will look at some choices we had to make in various projects. How the data is the first key to optimization. But also how to help the routing engine make the best of it. Which algorithms are best used for which use cases, and how fine tuning the database can help too.