minhh-nguyen/toydb-pams

How minhh-nguyen/toydb-pams works

toyDB is a distributed SQL database with a workload benchmark and experiment-analysis tools. The primary path is a SQL client sending a statement to a server, where a session parses and plans it, executes against transaction/catalog and storage layers, and returns results. Writes are coordinated through Raft replication; the README also documents workload generation and plotting/analysis of experiment data. Some subsystem internals were not sampled, so their detailed calls are omitted rather than inferred.

Main entry points: server.rs SQL and Raft server · session.rs SQL session · node.rs Raft node · executor.rs SQL executor

Components 25

Clients and service

SQL client src/bin/toysql.rs
SQL and Raft server src/server.rs

SQL processing

SQL session src/sql/execution/session.rs
Lexer and parser src/sql/parser/parser.rs
Planner and optimizer src/sql/planner/planner.rs
Execution plans src/sql/planner/plan.rs
SQL executor src/sql/execution/executor.rs
Aggregations and joins src/sql/execution
SQL types and schema src/sql/types
SQL engine src/sql/engine/engine.rs

Raft consensus

Raft node src/raft/node.rs
Raft log src/raft/log.rs
Raft messages src/raft/message.rs
Raft state src/raft/state.rs

Storage and encoding

MVCC transactions src/storage/mvcc.rs
Storage engines src/storage/engine.rs
BitCask and memory src/storage
Key and value encoding src/encoding/keycode.rs

Benchmark analysis

Workload benchmark src/bin/workload.rs
Experiment data analysis plot/config.py
Experiment charts plot

Other components

SQL user
Database cluster
CSV experiment data
Experiment analyst

Connections 25

SQL user → SQL client: submits SQL
SQL client → SQL and Raft server: sends requests
SQL and Raft server → SQL session: creates session
SQL session → Lexer and parser: parses SQL
SQL session → Planner and optimizer: builds plan
Planner and optimizer → SQL types and schema: uses schema
SQL session → SQL engine: opens transaction src/sql/execution/session.rs
SQL engine → MVCC transactions: uses MVCC src/sql/engine/engine.rs
Execution plans → SQL executor: executes plan
SQL executor → Aggregations and joins: delegates operators src/sql/execution/mod.rs
SQL executor → SQL engine: reads and writes
SQL and Raft server → Database cluster: serves SQL and Raft src/server.rs
SQL and Raft server → Raft node: routes Raft requests src/server.rs
Raft node → Raft log: manages log src/raft/node.rs
Raft node → Raft messages: exchanges messages src/raft/node.rs
Raft node → Raft state: applies state src/raft/node.rs
Raft log → Storage engines: persists log src/raft/log.rs
MVCC transactions → Storage engines: stores versions src/storage/mvcc.rs
Storage engines → BitCask and memory: implemented by src/storage/mvcc.rs
MVCC transactions → Key and value encoding: encodes keys src/storage/mvcc.rs
Workload benchmark → Database cluster: benchmarks
Database cluster → CSV experiment data: produces measurements
CSV experiment data → Experiment data analysis: loads CSV plot/config.py
Experiment data analysis → Experiment charts: supports charts plot/exp1/compare-latency.py
Experiment analyst → Experiment charts: generates charts

Drawn by GitDiagram on November 15, 2026 from the repository's file tree, README and source files. Read as Markdown ·