- Index is broken down, stored, and processed as collection of shards
- Shards are allocated to nodes and are searched independently
- Type is a mechanism to store different data in the same index
- Searching across types within index and between shards adds no overhead, almost
- Fields across types in an index must use consistent data types
- Score for search results calculated at the index level
- Mapping for document's properties are defined at the type level
- Search and aggregation can be executed at type, index, alias, multi-index, or at cluster levels