Relational is the new Big Data by Miguel Ángel Fajardo and Daniel Dominguez at Big Data Spain 2017

Relational databases were the persistence system of choice for decades, until the Web 2.0 in the 2000s required to process volumes of data so big it needed distributed systems running in parallel. A new type of databases (NoSQL) was adopted to solve this problem in different ways.

https://www.bigdataspain.org/2017/talk/relational-is-the-new-big-data

Big Data Spain 2017
16th - 17th November Kinépolis Madrid

Transcript

2. RELATIONAL is the new BIG DATA
3. Daniel Domínguez Head of Data Previously: CIEMAT / CERN @danieluchi01 Miguel Ángel Fajardo CTO Previously: EA Games, Gilt, Shutterstock @ma_bits
5. Distributed Processing Not Only SQL
9. A long time ago in a galaxy far, far away...
10. 1960s Information Management System (IMS) by IBM • Built for Saturn V moon rocket • Hierarchical, tree structure
12. 1970 Relational Model Paper • Base for IBM DB1 and DB2
14. 1980s-90s Development of RDBMS • Widely adopted • Models easy to define • ACID transactions • Clients for all stacks • ORMs
15. 2000s Web 2.0 • Large volume (petabytes) • Faster networks and devices • Systems must scale
16. Problems scaling Relational DBs • Sharding is hard • Maintaining transactions ACID is hard • Two-phase commit is hard • Parallelizing is hard
18. Distributed Processing Not Only SQL
19. Relational databases The CAP theorem
20. Key-value stores ◦ User session data ◦ Component configuration ◦ Cached data, fast access ◦ Complex queries ◦ Interconnected data
21. Column-oriented DB ◦ Real time analytics ◦ Facebook Messenger ◦ Queries against few rows ◦ Flexible data schemas ◦ Incremental data loads/deletes
22. ◦ Records with different fields ◦ Models with many layers ◦ Joins ◦ Flexible queries Document-oriented DB
23. Graph DB ◦ Routing ◦ Social networks ◦ Disease spreading ◦ Hard to do aggregates ◦ Analytics
24. No one magic database to rule them all • Each of them fits a small number of use cases • Often hard, complex and expensive to maintain • Specific query languages CQL
25. MEANWHILE IN THE RELATIONAL BATCAVE
26. 2010s Relational strikes back • Less structured data formats • Partitioning • Parallel execution • Sharding • C, A and P?
28. • ACID for queries going to a single shard • Open Source, DAAS • PostgreSQL extension • Interactive analytics • Multi-tenant • Fully ACID • Open Source • PostgreSQL fork • Scaling intensive • Multi-tenant
33. WHEN YOU HAVE A HAMMER...
34. Questions? tech.geoblink.com