The Computer Science Behind a Modern Distributed Data Store

Explore the computer science principles behind modern distributed data stores in this 55-minute conference talk. Delve into the challenges of developing resilient, distributed storage systems and learn about the key ingredients necessary for their implementation. Discover the significant progress made in both scientific understanding and practical applications of these data stores. Examine four modern approaches to open-source distributed data stores, including ArangoDB, Cassandra, Cockroach, and RethinkDB. Gain insights into consensus algorithms, distributed transactions, and query optimization. Cover topics such as Raft consensus, log-structured merge trees, hybrid logical clocks, and multiversion concurrency control. Understand the complexities of combining various components to achieve acceptable performance in distributed systems. Conclude with a Q&A session to address any remaining questions about this rapidly evolving field of computer science.

Introduction
Data stores are distributed
Consensus
The traditional solution
My advice
Raft
Home Protocol
Raft Demo
Sorting
Mergesort
Log structured merge trees
Log structured merge trees overview
Hybrid logical clocks overview
Hybrid logical clocks
Distributed transactions
Distributed systems
Multiversion concurrency control
Questions