Vertica Database Projections Types
What is a projection ?
is the Vertica phisical layer from wich data is queried.
projections are automaticly refreshed when new data is loaded or modified.
data is encoded and compressed in projections .
each projection stores a separate copy of data even if data is from the same table .
Projections are able to mantaing redundant copies on one or more cluster nodes of data
to obtain high availability .
Projection can also contain partitioned data .
Vertica can contain as many projection as needed to provide optimization for all queries.
some projection might contain the same data or similar data as other projections but may be sorted in a different way for optimization reasons.
Types of projections
Super Projection
contains all the column from a individual table.
they are contained automatilly after table are created and data was loaded.
Pre-Join Projection
it stores the result of join between Fact and a Dimension Table
created from denormalized tables .
join is autorefreshed as data is modified in the schema.
Query-Specific Projections
created by a subset of column from a table that has been specified for a specific query .
Buddy Projections
they contain the same columns and segmentation as another projection.
they located on multiple nodes to offer high availability
Way to create projections
the moment the table is loaded the super-projection is created.
un-optimized projections are created while initial load.
when optimized projections are crated the un-optimized projections
will be dropped .
manually , using the vsql CREATE PROJECTION command.
Share this with everybody
