Tarantool for banks
cover

Tarantool
for banks

With Tarantool data storage and processing, you can accelerate time-to-market of your products without drastically changing their architecture, create solutions for quick access to data, and create new banking services

Scroll

How does Tarantool help banks?

Increased availability and performance of key systems

Saving on licenses and equipment

Reduced number of systems to maintain

Integration of old protocols and systems into modern IT solutions

Fraud detection and prevention

Quick market launch of digital services

Tarantool use cases for banks

Data marts for core system hot data (CBS, CRM)

Omnichannel data marts for client profiles

Stream analytics in real time

Data marts for high-loaded services (e.g., account statements or investment business operations)

Real-time marketing and loyalty processing through online and offline channels

Calculation of aggregate function values for anti-fraud systems

Use case: increased availability
of core systems

Use case: increased availability
of core systems

Tarantool cache

Use case: omnichannel notifications

Use case: increased availability
of core systems

icon

Problem: as the client base and the number of client services grow, the number of requests to master services increases and the performance of information systems is reduced.

Task: ensure data availability of a major bank core-systems to reduce response time.

Use case: increased availability
of core systems

icon

Solution: a consistent cache for core-system data, which

  • supports read-through, write-through, and write-behind caching
  • guarantees fault-tolerance due to Tarantool cluster architecture
  • warms up quickly after restart due to WAL and snapshots
  • updates gradually through CDC
  • handles hundreds of thousands of updates per minute and more

Tarantool cache

Use case: omnichannel notifications

icon

Task: implement a single system for notification delivery through various channels.

Solution:

  • Tarantool is used as a storage for custom notification settings
  • A single system for managing notifications sent through different channels
  • Real-time omnichannel broadcast of notifications is possible
  • System features:

~20 million clients
~500 notifications per second on average
~2000 notifications per second at peak

icon icon icon
Scroll

Why implement a Tarantool-based solution

Use Tarantool within existing architecture as a middle layer with microservices built around it



Implement Tarantool in two months and reduce your project completion dates to weeks



Integrate Tarantool with business applications using Java and Go connectors

Use case: data mart
for real-time marketing

Use case: data mart
for real-time marketing

Real-time marketing

Use case: calculation of aggregate
function values for anti-fraud systems

Calculation of aggregate function values
for anti-fraud systems

Use case: data mart
for real-time marketing

icon

Problem: the data marts with marketing offers are different for each channel (mobile application, website, etc.). The channels get their data independently. The data marts of offers are inconsistent and out of sync. The data is stored in high-loaded core systems. A channel doesn’t react to events happening in other channels, while the large amount of user data (more than 10 TB) slows down information processing.

Task: implement a single system to store marketing offers and communications with clients from all channels.

Use case: data mart
for real-time marketing

icon

Solution:

  • A Tarantool-based data mart for all marketing offers that also stores all linked communications with the client
  • The data mart and communications are stored both in cache and in a persistent memory storage
  • The response delay is less than one millisecond due to in-memory technologies
  • The solution simplifies and automates preparing marketing offers, delivering them to Pull channels, collecting responses, sending Push communications, and doing analytics

Real-time marketing

Use case: calculation of aggregate
function values for anti-fraud systems

icon

Task: : implement transactional and analytical data processing for a fraud detection system operating in a bank. The system must analyze large amounts of data (more than 5 TB) in real time and detect suspicious transactions.

Solution:

  • The fraud detection system in real time makes a request to calculate aggregate function values through the HTTP API for each operation and rule
  • Tarantool calculates up to 2,500 aggregate functions with grouping by several sections in one rule (quantity, sum, average, minimum, maximum, median, etc.) reading legacy data up to 3-4 TB
  • The results are saved to a column storage for later use

Calculation of aggregate function values
for anti-fraud systems

icon icon icon
Scroll

Tarantool features

Based on a high-performance database. Can handle up to 1 million RPS

Data security through WALs, database snapshots, and a custom cluster management technology

Disaster-proof configurations with the Tarantool Clusters Federation tool

The Tarantool platform is certified by FSTEC of Russia

Tarantool products are included in the Russian register of domestic software

The products are developed by a Russian company and offer Russian vendor support