I'm currently working on a project where I have quite a bit of freedom in the structure and size of data. Now, I'm wondering if there's performance difference between the following two scenarios:
Assuming a two node setup with identical hardware
- 1 Index with 2 Shards, 1 replica
- Shard 1 contains data A
- Shard 2 contains data B
- 2 Indices with 1 Shard, 1 replica
- Index 1 contains data A
- Index 2 contains data B
Similar concepts can be thought up with 3 or 4 shards where the data would be split evenly in 3 or 4 shards on one index or 1 shard for each of the 3 or 4 indices.
In all honesty I'm expecting an answer to be that they have similar performance.
If that's the case I wonder if adding an extra node will have equal impact on both scenario's or will one scale better than the other?