Query hierarchical data the easy way, with CTEs
1 like222 views
The document discusses querying hierarchical data using Common Table Expressions (CTEs) in SQL, highlighting their availability in MariaDB since version 10.2. It covers both non-recursive and recursive CTEs, including syntax examples, use cases for improved readability, multiple references, and year-over-year comparisons. It also addresses the challenges of handling hierarchical data in SQL and explains how recursive CTEs can efficiently manage such structures.
Query hierarchical data the easy way, with CTEs
- 1. Querying Hierarchical Data with CTE ALEXANDER ALDEV DBaaS Architect MariaDB Corporation
- 2. Common Table Expressions ● SQL:1999 standard ● Temporary named result set ● Supported by: Oracle, MS SQL, PostgreSQL, SQLLite, MySQL, … ● Available in MariaDB since 10.2 ● Two types: ○ Non-recursive ○ Recursive
- 3. CTE Syntax WITH engineers AS ( SELECT * FROM employees WHERE dept = ‘engineering’ ) SELECT * FROM engineers ... CTE nameWITH keyword CTE body use in query
- 4. Similar to Derived Queries WITH engineers AS ( SELECT * FROM employees WHERE dept = ‘engineering’ ) SELECT * FROM engineers ... SELECT * FROM ( SELECT * FROM employees WHERE dept = ‘engineering’ ) engineers ...
- 5. Use Case: Readability WITH engineers AS ( SELECT * FROM employees WHERE dept = ‘engineering’ ), WITH eu_engineers AS AS ( SELECT * FROM engineers WHERE country IN (‘BG’,’FI’,’DE’) ) SELECT COUNT(*) FROM eu_engineers
- 6. Use Case: Multiple References to CTE WITH engineers AS ( SELECT * FROM employees WHERE dept = ‘engineering’ ) SELECT * FROM engineers e1 WHERE NOT EXISTS( SELECT 1 FROM engineers e2 WHERE e1.country = e2.country AND e1.first = e2.first )
- 7. Use Case: Year-over-Year comparison WITH annual_sales AS ( SELECT product, YEAR(invoice_date) AS year, SUM( amount ) AS total FROM invoices GROUP BY 1, 2 ) SELECT y1.year, y1.product, y1.total, y2.total last_yr_total FROM annual_sales y1 JOIN annual_sales y2 ON y1.year = y2.year + 1 AND y1.product = y2.product
- 8. Recursive CTE ● SQL does not handle hierarchical data well ● Trees in format (parent_id, child_id) ● Graphs in format (point 1, point 2) ● Recursive CTEs allow for hierarchical data to be handled ● Also useful for search algorithms where next iteration depends on previous
- 9. Recursive CTE Syntax WITH RECURSIVE ancestors AS ( SELECT * FROM family WHERE name = ‘Alex’ UNION ALL SELECT f.* FROM family f, ancestors WHERE f.parent_id = ancestors.id ) SELECT * FROM ancestors; SELECT * FROM engineers ... RECURSIVE keyword CTE anchor recursive use
- 10. How CTEs work? ● Base execution strategy is to materialize ● Optimizations possible for non-recursive CTE ● Recursive CTEs use this flow: 1. Evaluate anchor expression 2. Evaluate recursive expression -> new data 3. Append new data to result 4. If new data is not-empty goto 2
- 13. THANK YOU!
Editor's Notes
- #2: Alex: 20 yrs experience in DWH, analytics and a couple of business functions, currently leading MariaDB’s DBaaS architecture.
- #12: We’ll open two minikube consoles and do the following: Deploy a simple “Hello, world” NodeJS application. Use “docker build” and “kubectl create” Start a simple client that measures performance. Use “docker build” and “kubectl create” Scale the application from 1 to 3 and observe throughput. Use “kubectl get pods Fail a number of pods. Use “kubectl delete pod” Upgrade the application to v2.0. Use “kubectl set image”
- #13: We’ll open two minikube consoles and do the following: Deploy a simple “Hello, world” NodeJS application. Use “docker build” and “kubectl create” Start a simple client that measures performance. Use “docker build” and “kubectl create” Scale the application from 1 to 3 and observe throughput. Use “kubectl get pods Fail a number of pods. Use “kubectl delete pod” Upgrade the application to v2.0. Use “kubectl set image”
- #14: Last slide -- the remaining are backup and will be deleted