Revolut achieves 100x faster data queries after switch to Exasol analytics database

The UK fintech turned to Exasol to cope with an enormous growth of data while avoiding vendor lock-in

database futuristic technology
Getty Images

Revolut has increased the speed of the thousands of SQL queries it runs on a daily basis by 100 times after deploying the Exasol analytics database on the Google Cloud Platform.

The database has helped the fast-growing UK fintech business – which gained its full banking license in December – cope with the proliferation of data volumes and queries it has seen over the past couple of years.

In the four years since Revolut was founded the company has grown from a commission-free way of spending and transferring money abroad into a fully licensed provider of a wide variety of banking services, with over 6 million users and a $1.7 billion valuation.

That rapid growth has created an enormous volume of data that by last year had stretched the capacity of Revolut’s PostgreSQL relational database to its limits. Analytical queries were taking hours to run and the database couldn’t perform queries with complex joins and arbitrary fields.

Revolut tried to use the tools offered by its existing cloud provider, Google Cloud, but the pricing model proved too prohibitive for the level of data democratisation the business desired.

In Exasol it found a more affordable alternative: a consumption-based, in-memory relational database and analytics solution with on-premise, cloud or hybrid and multi-cloud deployment options.

Read next: Financial Times turns to graph technology to connect its vast array of data

Revolut chose to implement the multi-node Exasol cluster as a pay-as-you-go offering on Google Cloud infrastructure, without being locked into the Google Cloud environment. This allows Revolut to be flexible with its data strategy and extend its business to countries where Google Cloud is not directly available.

“Exasol had a unique combination of three things,” Demeter Sztanko, head of data engineering at Revolut, tells Computerworld. “First of all, it was the performance. Second, it was the amazing support and the lean, clear processes we received from them, so everything was organised really well. And the third was the attractive pricing combination.”

Deployment options

Sztanko says that the implementation of Exasol was simple enough that it took him just a week to set up a proof of concept database that was performing hundreds of times faster than Revolut’s previous system. Within three weeks, his team had completed a full deployment of the database.

He admits that they did encounter a few bugs after the initial implementation was completed and they started using Exasol more extensively, but was pleased by Exasol’s responses.

“Either a workaround was found immediately or they were fixed immediately,” he recalls. “That was a very positive customer experience.”

Now the dashboard solution is used by a total of around 600 employees every day, running around 100,000 SQL queries, performing tasks that range from checking funds in bank accounts to targeting deals at users by segmenting them into clusters.

The system has enabled them to complete queries that used to take hours within seconds and to maintain around 800 different dashboards and 12,000 graphs.

Read next: Oracle cofounder Larry Ellison slams AWS for security, cost and performance

Sztanko and his team now plan to use Exasol’s machining learning capabilities to help staff solve complex analytic challenges and allow the system to scale as demand continues to rise.

For other organisations evaluating their database options, Sztanko believes that Exasol is particularly well-suited to organisations that need to combine columns from multiple tables in their databases.

“If the nature of your data invokes cross-joining data a lot, you should definitely consider in-memory solutions – particularly Exasol,” he says. “This is where probably you could see the most of the performance gain compared to other analytical solutions.”

Copyright © 2019 IDG Communications, Inc.

How to supercharge Slack with ‘action’ apps