- Building: This is the building that should be constructed or on which construction work should happen. One building can have many construction work requests associated.
- Work: The construction work which should be done. One construction work entity belongs to exactly one building. One such an entity can have precisely one permit.
- Permit: The permit which was granted for requested construction work.
- Requester: A permit is requested by one requesting person. Such a requester can request multiple permits.
-
Building
- type : Enum
- address : String
- location : Geo
- neighbourhood: String
- works: Set (0..n)
-
Work
- type : Enum
- request_timestamp : Time
- requester: Person (1)
- building : Building (1)
- descr : Text
- permit: Permit (1)
-
Permit
- status : Enum
- timestamp : Time
- reason: Text
- work : Work (1)
-
Person
- first_name : String
- middle_name : String
- last_name : String
- birth_day : String
- address : String
- phone : String
- email : String
- work : Work (1..n)
- One index or multiple indices.
- most natural decision would be to create one index per
- RediSearch DOESN'T allow you to query across indices.
- Search queries are limited to the scope of a single index.
- One index per type: Each type is represented by exactly one index.
- Need multiple independent search queries to perform combined searches.
- Advantage - index as it is without data duplication to support combined queries.
- All types are covered by a single index: There is one search index that covers the fields of all types.
- Must create hashes containing all fields for query.
- Could lead to a serious duplication of data.
- Aggregates approach - nest objects within each other
- Create aggregate objects with multiple entities
- For example, the Permit entity can be nested in the Work entity as a 1-1 relationship, and the Work entity can be nested in the Building entity as a 1-n relationship.
- This nesting done with RedisJSON not Hashes
- RediSearch does not support indexing of multi-values (scalar values or objects nested in arrays), with the exception of TAGs.
Just a few examples on a few hash records
- add some permits hash records
redis-cli < add_hash_permits.txt- create index for records
redis-cli < permits_idx.txt- finally, run some queries out of queries.txt
redis-cli < add_json_permits.txt- create index for records
redis-cli < permits_json_idx.txt-
then, run some queries out of json_queries.txt
-
extra credit! run auto_complete example
redis-cli < auto_complete.txtmoodle data modelling basics