Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore a comprehensive database seminar talk on SplinterDB, a key-value store designed for modern storage devices. Delve into the SplinterDB story, learn about VSAN, and understand the external memory model. Examine comparison-based dictionaries and the use of hashing and filters. Investigate optimal data structures, including RoxDB, and explore various tree structures such as B-trees, Epsilon-B trees, and B-Epsilon trees. Discover the concept of buffer rewriting and how it improves lookups. Analyze the innovative Maplist and Maplet structures, addressing low memory concerns and redundant data issues. Gain insights into heap benchmarks, sequential insertions, data flushing techniques, and the benefits of parallel serialization. This in-depth presentation by Alex Conway from VMware Research SplinterDB offers valuable knowledge for database enthusiasts and professionals alike.
Syllabus
Introduction
SplinterDB Story
What is VSAN
External Memory Model
ComparisonBased Dictionaries
Using Hashing and Filters
Question Why is lookups not one read
Optimal Data Structures
Introducing RoxDB
Application Benchmark Suite
Data Structures
Btree
Epsilon
B Trees
B Epsilon Tree
Buffer Rewriting
Fixing Lookup
Using Filters
Fixing Lookups
The Problem with the Solution
What is a Maplist
Maplet
Low Memory
redundant data
multiple branches
mapplets
Heap
Benchmark
Sequential Insertions
Flushing Data
Benefits
Parallel Serialization
Taught by
CMU Database Group