LineairDB
LineairDB is a fast transactional key-value storage library. It provides transaction processing for multiple keys with strict serializability.
Features
- Keys and values are arbitrary byte arrays.
- The basic operations are read(key), write(key, value), ExecuteTransaction(procedure, callback).
- Changes in a transaction for multiple key-value pairs are made with atomicity and durability.
- Concurrent transactions are processed with strict serializability.
- In contended write-heavy workloads, high scalability for many-core CPUs is provided.
Notes
- LineairDB is not an SQL (Relational) database.
- There is no client-server support in the library (i.e., LineairDB is an embedded database).
Usage
#include <lineairdb/lineairdb.h>
int main() {
{
LineairDB::Database db;
LineairDB::TxStatus status;
// Execute: enqueue a transaction with an expected callback
db.ExecuteTransaction(
[](LineairDB::Transaction& tx) {
auto alice = tx.Read<int>("alice");
if (alice.has_value()) {
int alice = alice.value();
}
tx.Write<int>("bob", 1);
},
[&](LineairDB::TxStatus s) { status = s; });
// Fence: Block-wait until all running transactions are terminated
db.Fence();
// status == LineairDB::TxStatus::Committed;
}
}
Getting the Source
git clone --recurse-submodules https://github.com/lineairdb/lineairdb.git
Building
Quick start:
brew install boost # for OSX
mkdir -p build && cd build
cmake -DBUILD_SHARED_LIBS=ON -DCMAKE_BUILD_TYPE=Release .. && make && sudo make install
Then you can use LineairDB by including the header lineairdb/lineairdb.h
.
Compatibility
We have been tested LineairDB in the following environments:
- Apple clang version 11.0.3
- Clang >= 6 on Linux
- GCC >= 7.5
Documentation
The LineairDB library documentation is available.
You can also build it by simply running
doxygen
and open docs/build/html/index.html
.
Technical detail of concurrency control protocols such as SiloNWR is also available in the research paper at this link. The project roadmap is here.
Benchmark Results
The followings are our benchmark results. This is a modified benchmark of YCSB-A; unlike official YCSB in which a transaction operates a single key, each transaction operates on four keys in our benchmark.
This benchmark is executed in the following environments:
CPU | four Intel Xeon E7-8870 (total 144 logical cores) |
Memory | 1TB (no swap-out) |
YCSB Table size | 100K |
YCSB Record size | 8-bytes |
Epoch size | 1000ms |
Contention (θ) | 0.9 (highly contended) |
# of threads to process txns | 70 |
YCSB-A
SiloNWR, our novel concurrency control protocol, achieves excellent performance over 10 Million transaction per seconds (TPS) by omitting transactions without exclusive lockings. Note that YCSB-A has an operation ratio of Read 50% and (Blind) Write 50%; that is, this is a fairly favorable setting for NWR-protocols. If your use case is such a blind write-intensive, then LineairDB can be a great solution.
Contributing
This project welcomes contributions, issues, suggestions, and feature requests.
Feel free to check issues page.
Question & Discussion
If you have any questions, please feel free to ask on slack. Join to slack