Vertical sharding

simply split the database schema

• each group of related tables are put together on a different database By related meaning there are relationships between them (FK) or these group of tables often accessed together (JOINs) That’s the fancy word MicroServices simply as that. This can be done within the same node, or we can use different node for each database . Notice that each database now has a different schema
• break a large table into multiple tables instead of having Employee relation that has the following columns Id, name, … , Id_Picture let’s split the table into 2 tables Employee_Info, Employee_Assets now querying Employee_Info will be much much faster because now we ignore the assets (pictures) note that:

SELECT name
FROM Employee
WHERE Id > 10 and Id < 15

this query execution will include the blobs (pictures, large Json objects) internally before extracting the required columns out of the whole columns which is costly. This’s done within the same database instance, that’s perfect no need for extra nodes.

Main Goal of vertical sharding

For the table split it’s about fast retrieval no need to access the blob if we don’t need it. For the groups of tables split it’s about fast retrieval also, but now the load is split on different databases and then the system can accept more different unrelated queries and distribute them on the specific nodes holding the required intended data Maybe for example querying Employees needn’t to be on the same database as Beneficiaries