Choose a backend selection strategy
Compare round robin, least connections, weights, and hashing.
- Select a routing algorithm based on backend capacity and workload behavior.
Round robin cycles through available backends and works well when instances and requests are similar. Weighted round robin sends a larger share to higher-capacity instances. Least connections can help when request durations vary, though connection counts are only a proxy for work. Hash-based routing can keep a key on the same backend, but can create skew and complicates membership changes.
A policy is only as good as its inputs. Long-lived connections, uneven request cost, and stale health or load measurements can make a seemingly fair algorithm distribute actual work poorly.
Practice: pick the algorithm that survives
Try it
Up to 15 XPFind what works for each traffic pattern
Each scenario has a different answer. The bar shows the busiest moment for each backend, and the line marks what it can handle.
Try every algorithm, and work out why the others fail before moving on.
app-c has four times the capacity of the other two, and every request costs the same.
- app-abusiest: 4 of 3, handled 80
- app-bbusiest: 4 of 3, handled 80
- app-cbusiest: 4 of 12, handled 80
The bar is the busiest moment. The line marks what the backend can handle.
0 of 3 scenarios solved
Key takeaways
Use round robin for fairly uniform pools and workloads.
Weights represent known capacity differences.
No selection algorithm can compensate for misleading signals or severe key skew.
Lesson quiz
5 questions · pass with 4 correct · up to 50 XP
Passing this quiz completes the lesson and keeps your streak going. Questions you miss come back in review sessions later.
Questions about this lesson
Stuck? Ask. Figured something out? Share it. Explaining is one of the best ways to learn.
Loading posts…