HN Explorer

Ask HN: Petabyte Dictionary Lookup

2 points by flerovium ↗ HN
What is the cheapest and most efficient way to make a map from keys to values at very large scale, perhaps several petabytes?

When you query, the results are allowed to be a little stale.

It should be available to several boxes on a datacenter. Services that provide this are acceptable.

2 comments

[ 2.9 ms ] story [ 16.6 ms ] thread
icsa
Extendible hashing or Minimal perfect hashing
nikonyrh
Well the theory tells us that lookups are O(1) anyway, so the implementation doesn't matter ;)

Will the dataset change over time, or is it immutable? I have some thoughts on this, but I have no idea how it would scale to terabyte-scale and even further.