Cassandra Summit: C* Keys - Partitioning, Clustering, & Crossfit
3 likes2,113 views
This document summarizes key concepts for partitioning, clustering, and composite partition keys in Apache Cassandra. It begins with an introduction of the presenter and his company DataScale. It then defines partitioning as how nodes in a Cassandra cluster are assigned tokens to determine data distribution. The partition key is explained as the first column in the primary key that maps data to nodes, while clustering columns within each partition sort and organize the data. Composite partition keys using multiple columns for the token hash are also introduced. Examples of using clustering columns for hierarchical and time series data are provided.
Cassandra Summit: C* Keys - Partitioning, Clustering, & Crossfit
- 1. C* Keys: Partitioning, Clustering, & CrossFit Adam Hutson - Data Architect, DataScale Inc.
- 2. © DataStax, All Rights Reserved. Who am I & What do we do? 2 Adam Hutson Data Architect @ DataScale -> www.datascale.io DataStax MVP for Apache Cassandra DataScale provides hosted data platforms as a service Offering Cassandra & Spark, with more to come Currently hosted in Amazon & Azure
- 3. Overview
- 4. © DataStax, All Rights Reserved. 1 Why 2 Partition 3 Partition Key 4 Composite Partition Key 5 Clustering Columns 4
- 5. © DataStax, All Rights Reserved. Why give this presentation? Partitioning & Clustering should be the foundation. Too often glossed over. Has the biggest impact to performance of the cluster 5
- 6. Partition
- 7. © DataStax, All Rights Reserved. Partition Explained • Token values can range from -263 to 263-1. • Nodes in the cluster/ring are assigned a single token. • A node is responsible for the token value and expands to the previous node’s token. • A Partitioner decides where a partition key maps onto the cluster/ring. 7 Node #3 is responsible for tokens from -1844674407370955162 to -5534023222112865485
- 8. © DataStax, All Rights Reserved. Partition Explained 8
- 10. © DataStax, All Rights Reserved. Partition Key Explained The Partition Key is: • responsible for distribution of data amongst the nodes • the first column defined in the PRIMARY KEY 10
- 11. © DataStax, All Rights Reserved. Partition Key Explained 11
- 12. © DataStax, All Rights Reserved. Partition Key Explained 12
- 14. © DataStax, All Rights Reserved. Composite Partition Key Explained Using multiple columns for the token hash value. 14
- 15. © DataStax, All Rights Reserved. Composite Partition Key Explained 15
- 16. © DataStax, All Rights Reserved. Composite Partition Key Explained 16
- 18. © DataStax, All Rights Reserved. Clustering Columns Explained Clustering Columns are: • responsible for sorting within the partition • any column added to the Primary Key, past the first column 18
- 19. © DataStax, All Rights Reserved. Clustering Columns Explained Can be used for Hierarchical structured data. 19
- 20. © DataStax, All Rights Reserved. Clustering Columns Explained Can be used for Time Series structured data. CREATE TABLE member_log ( member text, workout_date timestamp, workout_duration text, PRIMARY KEY (member, workout_date) ) WITH CLUSTERING ORDER BY (workout_date DESC); 20
- 21. © DataStax, All Rights Reserved. Clustering Columns Explained 21
- 22. Thank You! Questions? Adam Hutson @AdamHutson [email protected] @DataScaleInc