AWS - Redshift

Posted on |

Reference

https://youtu.be/lj8oaSpCFTc

Terminology

  • massively parallel, share Nothing Columnar architecture

Best Practices: Encoding & Compression

https://youtu.be/lj8oaSpCFTc?t=657

  • Use AZt4

Basics

  • blocks (1MB immutable block encoded with 1 encoding)
  • zone maps
  • sort key

https://youtu.be/lj8oaSpCFTc?t=787

Best Practices: Sort Keys

https://youtu.be/lj8oaSpCFTc?t=941

  • Compound key: Lowest cardinality columns first
  • Use script to help find the sort key
  • Define sort key on large table, four or less columns

Best Practice: Materialize columns

https://youtu.be/lj8oaSpCFTc?t=1001

  • Frequently filtered and unchanging dimension values should be materialized within fact tables;

Basics: Slice, Data Distribution

https://youtu.be/lj8oaSpCFTc?t=1114

Best practices: table design summary

https://youtu.be/lj8oaSpCFTc?t=1455

Reward Makes Perfect
0%