What is FnckSQL

FnckSQL is a high-performance SQL database that can be embedded in Rust code (based on RocksDB by default), making it possible to call SQL just like calling a function. It supports most of the syntax of SQL 2016.

Quick Started

Tips: Install rust toolchain and llvm first.

Clone the repository

git clone https://github.com/KipData/FnckSQL.git

Using FnckSQL in code

let fnck_sql = DataBaseBuilder::path("./data").build()?;
let tuples = fnck_sql.run("select * from t1")?;

TPCC

run cargo run -p tpcc --release to run tpcc

• i9-13900HX
• 32.0 GB
• YMTC PC411-1024GB-B
• Tips: TPCC currently only supports single thread

<90th Percentile RT (MaxRT)>
   New-Order : 0.005  (0.007)
     Payment : 0.084  (0.141)
Order-Status : 0.492  (0.575)
    Delivery : 6.109  (6.473)
 Stock-Level : 0.001  (0.001)
<TpmC>
89.9205557572134 Tpmc

PG Wire Service

run cargo run --features="net" to start server then use psql to enter sql

Storage Support:

• RocksDB

Docker

Pull Image

docker pull kould23333/fncksql:latest

Build From Source

git clone https://github.com/KipData/FnckSQL.git
cd FnckSQL
docker build -t kould23333/fncksql:latest .

Run

We installed the psql tool in the image for easy debug.

You can use psql -h 127.0.0.1 -p 5432 to do this.

docker run -d \
--name=fncksql \
-p 5432:5432 \
--restart=always \
-v fncksql-data:/fnck_sql/fncksql_data \
-v /etc/localtime:/etc/localtime:ro \
kould23333/fncksql:latest

Features

• ORM Mapping: features = ["macros"]

#[derive(Default, Debug, PartialEq)]
struct MyStruct {
  c1: i32,
  c2: String,
}

implement_from_tuple!(
    MyStruct, (
        c1: i32 => |inner: &mut MyStruct, value| {
            if let DataValue::Int32(Some(val)) = value {
                inner.c1 = val;
            }
        },
        c2: String => |inner: &mut MyStruct, value| {
            if let DataValue::Utf8(Some(val)) = value {
                inner.c2 = val;
            }
        }
    )
);

• User-Defined Function: features = ["macros"]

scala_function!(TestFunction::test(LogicalType::Integer, LogicalType::Integer) -> LogicalType::Integer => |v1: DataValue, v2: DataValue| {
    let plus_binary_evaluator = EvaluatorFactory::binary_create(LogicalType::Integer, BinaryOperator::Plus)?;
    let value = plus_binary_evaluator.binary_eval(&v1, &v2);

    let plus_unary_evaluator = EvaluatorFactory::unary_create(LogicalType::Integer, UnaryOperator::Minus)?;
    Ok(plus_unary_evaluator.unary_eval(&value))
});

let fnck_sql = DataBaseBuilder::path("./data")
    .register_scala_function(TestFunction::new())
    .build()?;

• User-Defined Table Function: features = ["macros"]

table_function!(MyTableFunction::test_numbers(LogicalType::Integer) -> [c1: LogicalType::Integer, c2: LogicalType::Integer] => (|v1: DataValue| {
    let num = v1.i32().unwrap();

    Ok(Box::new((0..num)
        .into_iter()
        .map(|i| Ok(Tuple {
            id: None,
            values: vec![
                DataValue::Int32(Some(i)),
                DataValue::Int32(Some(i)),
            ]
        }))) as Box<dyn Iterator<Item = Result<Tuple, DatabaseError>>>)
}));
let fnck_sql = DataBaseBuilder::path("./data")
  .register_table_function(MyTableFunction::new())
  .build()?;

• Optimizer
  • RBO
  • CBO based on RBO(Physical Selection)
• Execute
  • Volcano
• MVCC Transaction
  • Optimistic
• Field options
  • [not] null
  • unique
  • primary key
• SQL where options
  • is [not] null
  • [not] like
  • [not] in
• Supports index type
  • PrimaryKey
  • Unique
  • Normal
  • Composite
• Supports multiple primary key types
  • Tinyint
  • UTinyint
  • Smallint
  • USmallint
  • Integer
  • UInteger
  • Bigint
  • UBigint
  • Char
  • Varchar
• DDL
  • Begin (Server only)
  • Commit (Server only)
  • Rollback (Server only)
  • Create
    • Table
    • Index: Unique\Normal\Composite
    • View
  • Drop
    • Table
    • Index
    • View
  • Alert
    • Add Column
    • Drop Column
  • Truncate
• DQL
  • Select
    • SeqScan
    • IndexScan
  • Where
  • Distinct
  • Alias
  • Aggregation: count()/sum()/avg()/min()/max()
  • SubQuery[select/from/where]
  • Join: Inner/Left/Right/Full/Cross (Natural\Using)
  • Group By
  • Having
  • Order By
  • Limit
  • Show Tables
  • Explain
  • Describe
  • Union
• DML
  • Insert
  • Insert Overwrite
  • Update
  • Delete
  • Analyze
  • Copy To
  • Copy From
• DataTypes
  • Invalid
  • SqlNull
  • Boolean
  • Tinyint
  • UTinyint
  • Smallint
  • USmallint
  • Integer
  • UInteger
  • Bigint
  • UBigint
  • Float
  • Double
  • Char
  • Varchar
  • Date
  • DateTime
  • Time
  • Tuple

Roadmap

• SQL 2016

License

FnckSQL uses the Apache 2.0 license to strike a balance between open contributions and allowing you to use the software however you want.

Contributors

Thanks For

• Fedomn/sqlrs: Main reference materials, Optimizer and Executor all refer to the design of sqlrs
• systemxlabs/bustubx
• duckdb/duckdb