Performance for Djangos
Django performance work focuses on reducing database queries and repeated computation. Query optimisation uses select_related and prefetch_related to avoid the N+1 problem, indexes and only() or values() for lighter reads, while caching stores expensive results using backends such as Redis or Memcached at the view, template fragment or low-level API level.
Stage: Data · Part of the Django roadmap · Last reviewed 19 November 2026
What to learn in Performance
- Spot an N+1 query with the debug toolbar or query logging
- Fix N+1 using select_related and prefetch_related
- Add database indexes with Meta.indexes
- Cache a view and a fragment with the cache framework
- Invalidate cached values when the underlying data changes
- Use only(), defer() and values() to limit loaded columns
Performance interview questions
What is the N+1 query problem and how do you fix it in Django?
It happens when you load a list of objects and then access a related object on each, causing one extra query per row. Use select_related for foreign keys and one-to-one, or prefetch_related for many-to-many and reverse relations, to load them together.
When should you use select_related versus prefetch_related?
select_related performs a SQL join and suits single-valued relations such as ForeignKey. prefetch_related runs a separate query and joins in Python, suiting many-valued relations such as ManyToMany or reverse foreign keys.
Where Performance fits in the Django roadmap
Performance belongs to the Data stage (intermediate level), which takes about 20 hours. ORM and migrations.
Other units in this stage
Back to the full Django roadmap