Expert Python Programming - Third Edition

Using architectural trade-offs

When your code can no longer be improved by reducing the complexity or choosing a proper data structure, a good approach may be to consider a trade-off. If we review users' problems and define what is really important to them, we can often relax some of the application's requirements. Performance can often be improved by doing the following:

• Replacing exact solution algorithms with heuristics and approximation algorithms
• Deferring some work to delayed task queues
• Using probabilistic data structures

Let's move on and take a look at these improvement methods.