Non-Uniform Distribution
Hash functions are deterministic but not perfectly uniform when applied to a small number of inputs. With only 3 or 4 servers, you might get unlucky placements:
hash("Server A") → position 10
hash("Server B") → position 12
hash("Server C") → position 90
Server C now owns the arc from position 13 to 90 — 78% of the ring. It will receive roughly 78% of all traffic while Servers A and B share the remaining 22%. This is called data skew or a hotspot.
Heterogeneous Hardware
A second problem: standard consistent hashing treats all servers as equal. If Server C is a 64-core machine with 512 GB RAM and Server A is a 4-core machine with 16 GB RAM, giving them equal ring presence wastes Server C's capacity and overloads Server A.
Both problems — skew from unlucky hash placement and wasted heterogeneous capacity — are solved by Virtual Nodes.