Tarantool Column Store

Main article: Database Management System (DMS)

2024: Launch of Hybrid Data Processing Solution

VK has launched a real-time hybrid processing solution. The company announced this on February 21, 2024.

Tarantool Column Store is a column DBMS designed for hybrid processing of streaming data in RAM. The solution instantly analyzes the most up-to-date transaction information and reduces the cost of migrating arrays to storage.

Traditionally, in the IT infrastructure of a business, transactional and analytical systems are separated from each other. Large data streams have to be transferred to a storefront or store. After that, information can be analyzed without slowing down operational processes. This reduces the relevance of data and received insights, as well as increases business costs. In-memory DBMS Tarantool Column Store in moments provides hybrid processing of transactions and analytical requests in one system, answering modern calls. So, some AI tasks require a data processing speed of up to 1.5 million records per second on one processor core, "said Nikolai Karlov, director of innovative projects at VK Tech.

Tarantool Column Store helps the business work with financial and production data, consolidate it, speed up analytical queries and generate real-time reporting. The DBMS flexibly scales computing resources, increases the availability of information, and provides high processing performance.

For example, DBMS financial institutions will help improve the quality and speed up the operation of anti-fraud systems, reducing the delay in detecting suspicious transactions to almost zero. The solution also allows you to increase the speed and accuracy of credit issuance services thanks to an instant assessment of the risk of non-repayment.

Tarantool Column Store is a hybrid DBMS with support for both column and line storage and data processing. The product has a high-performance vectorized multi-threaded SQL query engine and integrates with business applications using JDBC/ODBC/ADBC and HTTP. DBMS can be scaled by clustering, replication, and charring. Built-in mechanisms for compressing data in RAM, as well as the ability to use DSS for archiving, allow you to work with data arrays that do not fit into RAM. The solution is compatible with the Russian operating systems Astra Linux, RED OS and Alt.