Explore the evolution and design principles of distributed SQL databases in this 42-minute conference talk from Strange Loop. Delve into the history of database technologies, from primitive systems to modern distributed SQL engines. Discover how these new systems combine the transactional consistency of traditional relational databases with the scalability and availability of NoSQL solutions. Examine the key design choices behind CockroachDB, an open-source distributed SQL database, including dynamic sharding, consensus protocols, and lock-free transactions. Learn about data distribution techniques, replication strategies, and the intricacies of distributed consensus using Raft. Gain insights into transaction management, conflict resolution, and the challenges of maintaining consistency in a distributed environment. Understand the motivations behind the development of distributed SQL databases and their potential impact on future database design and implementation.
Overview
Syllabus
Introduction
History
Primitive databases
The relational model
Objectoriented databases
Database sharding
No sequel databases
Distributed sequel databases
Why these systems are being built
The two primary systems
The question of how
Disclaimer
Data Distribution
Order Preserving
Range Indexing
Range Scan
When to Split
When to Up Replicate
Primary Secondary Replication
Eventually Consistent
Distributed Consensus
Raft
When things dont go well
Engineering complications
Transactions
Consistency
Atomicity
MVCC
No sequel transactions
Distributed sequel
Starting a transaction
Conflicts
Write Conflict
Detect Conflict
Push or abort
Intent
Summary
Outro
Taught by
Strange Loop Conference
Tags
Related Courses
-
Raft - The Understandable Distributed Protocol
-
Anatomy of a Distributed SQL Database - YugabyteDB
-
Distributed Principles in a Cloud-Native Database - CockroachDB Deep Dive
-
Don't Give Up on Serializability Just Yet
-
Spanner - Google's Distributed Database
-
Building Blocks of Distributed Systems - Parts 1 & 2
