Course contentsShow
← System Design
Lesson 1468 of 1,91936. Data Partitioning Strategies in DepthFree lesson

Bounded Loads Extension

Preventing hotspots by limiting maximum load per node and redirecting excess keys to alternate nodes.

Bounded Loads Extension

What you'll learn: How to prevent any single node from becoming overloaded by capping its maximum load and gracefully redirecting excess traffic to other nodes.


The Hotspot Problem in Consistent Hashing

Even with virtual nodes, consistent hashing can create hotspots—nodes receiving disproportionately high traffic due to popular keys or uneven access patterns. A single viral post, trending product, or celebrity profile can overwhelm the node responsible for that key, degrading performance for everyone.

How Bounded Loads Works

The bounded loads extension adds a simple rule: no node should handle more than a maximum threshold of the average load.

Here's the algorithm:

  1. Calculate the average load across all nodes
  2. Set a bound (typically 1.25× or 1.5× the average)
  3. When assigning a key, walk clockwise around the hash ring as usual
  4. Skip overloaded nodes (those at or above the bound)
  5. Assign to the first available node that's under the threshold

Practical Example

Imagine 4 nodes each should handle ~25% of keys on average:

  • Average load = 25%
  • Bound = 1.25 × 25% = 31.25%
  • If Node A hits 31.25%, new keys skip it and go to Node B
  • This prevents Node A from reaching 50% while others sit at 15%

The Safety Net Analogy

Think of it like a restaurant with overflow seating: if the main dining room (primary node) reaches capacity, the host directs guests to the patio (alternate node) rather than cramming everyone into one space.

Key Takeaway:
Bounded loads prevents hotspots by capping each node's maximum load at a multiple of the average, redirecting excess keys to the next available node on the hash ring.