Modeling Redis Hashes as Tables
The provider can be configured to shape the discovered metadata.
Use the DefineTables, TablePattern, and PatternSeparator connection properties to customize how tables and columns are inferred from the Redis key store.
Presume the following hashes have been created in the Redis server (either with redis-cli or the RunCommand storec procedure).
> hmset user:1000 name "John Smith" email "[email protected]" password "s3cret" OK > hmset user:1001 name "Mary Jones" email "[email protected]" password "hidden" OK > hmset user:1002 name "Sally Brown" email "[email protected]" password "p4ssw0rd" OK > hmset customer:200 name "John Smith" account "123456" balance "543.21" OK > hmset customer:201 name "Mary Jones" account "123457" balance "654.32" OK > hmset customer:202 name "Sally Brown" account "123458" balance "765.43" OK
When these properties are used to define the provider's behavior, the Redis keys will be pivoted, so that each Redis key that matches the pattern in the definition is represented as a single row in the table. Each value associated with that Redis key becomes a column for the table.
DefineTables Property
The DefineTables connection property allows you to explicitly define the names of the tables that will appear. To do so, set the property to a comma-separated string of name-value pairs, where the name is the name of the table and the value is the pattern used to assign Redis keys to that table.
The provider aggregates all of the Redis keys that match the specified patterns.
DefineTables=Users=user:*,Customers=customer:*;
With the property set as above, the Users and Customers tables will be exposed. If you were to query the tables, you would see the following results:
SELECT * FROM Users
RedisKey | name | password | |
user:1000 | John Smith | [email protected] | s3cret |
user:1001 | Mary Jones | [email protected] | hidden |
user:1002 | Sally Brown | sally.b@example | p4ssw0rd |
SELECT * FROM Customers
RedisKey | name | account | balance |
customer:200 | John Smith | 123456 | 543.21 |
customer:201 | Mary Jones | 123456 | 654.32 |
customer:202 | Sally Brown | 123456 | 765.43 |
TablePattern Property
The TablePattern connection property allows you to define the separator(s) that determine how the provider defines tables. For the Redis keys described above, "user" and "customer" would be defined as tables if the separator is set to ":" since the unique piece of each Redis key appears after the ":". If you have a need to structure the tables differently, to drill down further, you can include multiple instances of the separator. Set the property to a pattern that includes the separator(s) needed to define your table structure. (Below is the default value.)
You can also manually specify the pattern separator indepently from the TablePattern using the PatternSeparator property.
TablePattern=*:*;
With the property set as above, the user and customer tables will be exposed. If you were to query the tables, you would see the following results:
SELECT * FROM user
RedisKey | name | password | |
user:1000 | John Smith | [email protected] | s3cret |
user:1001 | Mary Jones | [email protected] | hidden |
user:1002 | Sally Brown | sally.b@example | p4ssw0rd |
SELECT * FROM customer
RedisKey | name | account | balance |
customer:200 | John Smith | 123456 | 543.21 |
customer:201 | Mary Jones | 123456 | 654.32 |
customer:202 | Sally Brown | 123456 | 765.43 |