Hi guys.
About defining the storage format (columnstore or rowstore) and how to fragment tables or not, I would like to know if the classifications I put in each of these 6 little examples are correct.
Example: 01
Storage: columnstore
Shards: randomly distributed values
CREATE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL,
KEY (S_Nation) USING CLUSTERED COLUMNSTORE);
Example: 02
Storage: columnstore
Shards: coordinated distributed values
CREATE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL,
SHARD KEY (S_Nation),
KEY (S_Nation) USING CLUSTERED COLUMNSTORE);
Example: 03
Storage: columnstore
Shards: reference table (full replication on all nodes)
CREATE REFERENCE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL,
KEY (S_Nation) USING CLUSTERED COLUMNSTORE);
Example: 04
Storage: rowstore
Shards: randomly distributed values
CREATE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL);
Example: 05
Storage: rowstore
Shards: coordinated distributed values
CREATE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL,
SHARD KEY (S_Nation));
Example: 06
Storage: rowstore
Shards: reference table (full replication on all nodes)
CREATE REFERENCE TABLE supplier(
S_SuppKey INT(11) NOT NULL,
S_Name VARCHAR(25) NOT NULL,
S_Nation VARCHAR(10) NOT NULL);