KaKao

Kakao opens codes for open communication between internal and external developers, so that they can learn and grow together.

 

Kakao’s open source 

Buffalo: Buffalo is a fast, scalable production-ready open source project for recommendation systems. Buffalo effectively uses system resources to provide high performance in low-spec systems.

 

khaiii: The name “khaiii” is taken from the first letters of "Kakao Hangul Analyzer III", and is the third morphological analyzer developed by Kakao. khaiii is aptly named after its predecessor, Kakao’s second morphological analyzer named “dha2 (Daumkakao Hangul Analyzer 2)”.

 

Apache S2Graph: S2Graph is a graph database designed to process transactional graphs at scale. The REST API can be used to store, manage and query relational information in a fully asynchronous and non-blocking manner using edge and vertex representations.

 

TOROS N2: N2 is a Nearest Neighbor library written in C++ (including Python/Go bindings). N2 offers faster search speed than other implementations when modeling large datasets, and supports multi-core CPU for index building.

 

HBase Tools: HBase Tools is a collection of tools that was created while Kakao operated large-scale HBase clusters. HBase builds are provided for each version, which can be used as-is, or as a starting point to build personalized HBase tools.

 

HBase Region Inspector: HBase Region Inspector is a practical tool that provides a visual dashboard of data that is scattered about HBase’s many regions. Kakao uses this useful tool to operate large HBase clusters.

 

HBase Packet Inspector: HBase Packet Inspector is a tool that analyzes HBase region server’s network packets to extract request and responses. This helpful tool provides detailed information that is not available in previous monitoring tools.

 

CMUX: CMUX provides interactive command line interface tools needed to manage Cloudera Manager-based Hadoop clusters. CMUX ideas area always available for reference. Use CMUX and see your command lines take flight.

 

MRTE-Collector: MySQL Realtime Traffic Emulator (MRTE) is comprised of two tools: 1) MRTE-Collector, which collects the traffic of MySQL servers in the actual service, and 2) MRTE-Player, which reproduces the collected data in the test MySQL server.

 

ADT - Almighty Data Transmitter: ADT’s original purpose was to redistribute shards when the shard construction or sharding rules changed. However, ADT is transforming into a powerful data transmission tool (that combines CDC and ETL) that collects data in MySQL and sends the data to another MySQL.