One of the underlying assumption/requirement for Consistent Hashing to work well is that the Hash Function will produce uniformly distributed hash over a specified range (say 1 … 256). The question then is, while Consistent Hashing would be able to help with adding more servers (storage capacity) with minimal disruption, why should Consistent Hashing be necessary for achieving uniform load distribution?
@CodeRank In cases where one of the servers fails and leaves the system, the load on the rest of the servers becomes non-uniform. The non-uniform data distribution might create some data hotspots across our servers, which will make the system inefficient.
For a more clear picture and understanding, have a look at https://www.acodersjourney.com/system-design-interview-consistent-hashing/
1 Like