Concept
What is vectorized execution?
Understand vectorized execution as batch processing instead of row-at-a-time work, especially in OLAP systems.
CORE CONCEPTS
Core concepts
Explain shared data-system principles with intuitive models and minimal examples.
This section connects the common foundations across projects: storage, execution, scheduling, data distribution, and query optimization. Understand the concept first, then return to the implementation.
Concept
What is shuffle?
Why distributed systems redistribute data, and where shuffle cost comes from.
Compare
Spark vs Ray: batch processing and distributed applications
Understand Spark and Ray through computation models, scheduling, data work, and ecosystem boundaries.
Compare
Spark vs Daft: different data-processing models
Compare Spark and Daft across data abstractions, physical plans, ecosystem, and fit.
Docs
Spark
A learning path through execution, shuffle, Spark SQL, and streaming.
Concept
Row storage and columnar storage
Understand row and column storage through access patterns, compression, scans, and write cost.
Docs
Ray
Understand Ray’s distributed application model through Tasks, Actors, and scheduling.
Docs
Doris
Notes on MPP analytics, partitioning, bucketing, table models, and acceleration.
Docs
Daft
Notes connecting DataFrames, physical plans, and distributed data processing.
Compare
ClickHouse vs Doris: choosing an analytical database
A comparison framework for ClickHouse and Doris across data layout, execution, ecosystem, and operations.
Docs
ClickHouse
A practical path through MergeTree, ordering keys, partitions, and query execution.