Tables in Redshift

Introduction to Redshift

Jason Myers

Principal Engineer

Creating Tables

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    'organization_id' VARCHAR(31),
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
);

Distributing data

Redshift Cluster

Distributed across compute nodes
Uses Redshift internal row id, DISTKEY, or PRIMARY KEY
Several distribution styles

Distribution styles

Name	Description	Usage
ALL	Entire table on every node	Small fact lookup tables we need often in joins
KEY	Distributed by data in the DISTKEY column	When we aggregate or join by DISTKEY
EVEN	In turn distribution across nodes by row	Large tables with that don't have keys
AUTO	Uses ALL for small tables. Key as it grows if there are suitable DISTKEYS falling back to EVEN	default

DISTKEY

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    'organization_id' VARCHAR(31),
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
)
-- Sets the data distribution key 
-- to organization_id
DISTKEY(organization_id);

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    -- Sets organization_id as the data 
    -- distribution key 
    'organization_id' VARCHAR(31) DISTKEY,
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
);

Setting the distribution style

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    'organization_id' VARCHAR(31),
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
)
DISTKEY(organization_id)
-- Sets the distribution style to key
DISTSTYLE KEY;

SORTKEY

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    'organization_id' VARCHAR(31),
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
)
DISTKEY(organization_id)
DISTSTYLE KEY
-- Sets the data sort key 
-- to fk_monitoringlocation
SORTKEY(fk_monitoringlocation);

Controls the storage order on disk
Amplifies the power of predicate-pushdown
Can have multiple

Defining multiple SORTKEYs

CREATE TABLE IDAHO_SITE_ID
(
    'pk_siteid' INTEGER PRIMARY KEY,
    'fk_monitoringlocation' INTEGER,
    'organization_id' VARCHAR(31),
    'organizationformalname' VARCHAR(68),
    'organization' VARCHAR(16)
)
-- Sets fk_monitoringlocation, organization_id as compound sort keys
COMPOUND SORTKEY(fk_monitoringlocation, organization_id);

Viewing column DISTKEY and SORTKEY status

-- View the dist and sortkey
-- status of a column
SELECT column_name, 
       distkey, 
       sortkey
  FROM SVV_REDSHIFT_COLUMNS
 -- Only in the spectrumdb schema
 WHERE schema_name = 'spectrumdb'
   -- For the ecommerce_sales table
   AND table_name = 'ecommerce_sales';

Results

   column   | distkey | sortkey
============|=========|========
year_qtr    | t       | 1
total_sales | f       | 2
ecom_sales  | f       | 0

Distkey Column
- t for True - F for False
SORTKEY Column
- # is position in sort key, 0 is not part of sortkey

Viewing distribution style

SVV_TABLE_INFO
Table details that affect query performance
- distribution style
- distribution skew
- table size
- sortkeys
- sortkey skew

table          | encoded | diststyle       | sortkey1     | skew_sortkey1 | skew_rows
===============|=========|=================|==============|===============|===========
ecommerce_sales| N       | KEY(year_qtr)   | year_qtr     |               |          
date           | N       | ALL             | dateid       |          1.00 |

Viewing distribution style continued

SELECT table
       diststyle 
  FROM SVV_TABLE_INFO
 WHERE schema like 'spectrumdb';

table           | diststyle 
================|============== 
ecommerce_sales | KEY(year_qtr)

Let's practice!

Introduction to Redshift