up Gazprombank’s
internal systems
by a factor
of 50 with Tarantool

Gazprombank is the third largest Russian bank by assets and the fourth largest one in terms of profit. The bank serves more than 5 million customers.


Tasks and requirements

Information about banking cards, deposits, unallocated metal accounts, and other banking products is stored in a core banking system (CBS) that was implemented in 2006. Extensive usage of mobile applications and client portals has created the pressure of multiple simultaneous requests that the system could hardly withstand.

Relational databases that are the basis of modern CBSs could not operate fast enough for both customers and bank employees. Therefore, the bank decided to implement an operational cache with the following characteristics:

  • No more than a 5-second delay from the master system.

  • Smart cache that monitors data consistency, expiration, and lagging behind the master system.
  • Ability to select associated data with a  single query using a cluster query language in consumer systems.
  • 200 ms SLA for reads even for complex queries to associated data.

The bank team was looking for a vendor that would be ready to create turnkey solutions, be responsible for integrations with other systems, work with the bank analysts, and operate in Agile mode.






In mid-2019, Gazprombank launched a project to develop and implement a system based on Tarantool Data Grid. The solution caches data from different banking information systems and ensures the desired performance.

All the data, both basic and operational, is stored in RAM. This ensures the necessary level of performance for any workload: up to hundreds of thousands of requests per second per an instance.

Data changes are immediately available to all data consumers, so the solution can be classified as a real-time system. It provides data with a frequency of about 1500 RPS and a response time of less than 100 ms.

The project consisted of several phases. The first phase involved implementing the mechanism of loading bank card information from the CBS into the system. This phase was necessary to check and debug the system in production in terms of integration and periodical data loading.




At the second phase, the bank implemented card product data delivery from the cache to the unified front end.

At the final phase, full data set upload was implemented in the unified front end: deposits, accounts, unallocated metal accounts, and loans could now be requested from the system.

With Tarantool Data Grid, Gazprombank extended the functionality of customer service channels without changing the existing banking system. Thus, the bank managed to create a solution with easily extensible functionality while maintaining high performance and full data reliability.

icon icon



The solution can handle more than 100,000 requests per second from other banking systems.

100 ms

Data is provided with delays less than 100 ms, that is, the solution can be classified as a real-time system.

Horizontal scalability takes all questions about the data limit off the table—the limit simply doesn’t exist.

Get a consultation

Order a demo

Thank you for your request

Tarantool experts will
contact you shortly