I have been reading up on distributed hashing. I learnt that consistent hashing is used for distributing the keys among cache machines. I also learnt that, a key is duplicated on mutiple caches to handle failure of cache hosts. But what I have come across on memcached doesn't seem to be in alignment with all this. I read that all cache nodes are independent of each other and that if a cache goes down, requests go to DB. Theres no mention of cache miss on a host resulting in the host directing the request to another host which could either be holding the key or is nearer to the key.
Can you please tell me how these two fit together? Is memcached a very preliminary form of distributed hashing which doesnt have much sophistication?