Optimising For BFCM Success: A Behind-the-Scenes Look at How Loqate Handled 1 Billion Transactions

David Starling, Software Engineer Manager

Black Friday and Cyber Monday is a high-stakes period for eCommerce. With such a high volume of transactions, it is crucial to collect accurate addresses for customers at the point of capture. Poor quality address data during these high traffic days can be extremely costly for eCommerce businesses and have significant negative impact on their bottom line.  

At Loqate, we obsess about continuously sustaining high-levels of reliability at ever greater scale, so our customers can feel confident even while they are pushing their infrastructure to its limits in response to the surge in consumer demand. 

Loqate processed more than 1 billion requests to our API – the largest ever four-day period for Loqate – with greater than 99.99% availability and a peak of 5,600 requests per second. This level of reliability ensured that large companies such as ASOS, Fanatics and Sephora maximised their deliverability and revenue potential throughout the most active shopping days of the year. 

Loqate’s BFCM request volume has increased more than 100x over the last ten years. Maintaining this level of growth year-on-year, requires great software and also great infrastructure. 

In order to build such a robust platform, we need something flexible, scalable and secure. These platform qualities enable our team to ship with confidence and iterate efficiently. 

Flexible, Scalable, and Secure  

Loqate engineering relies on Google Kubernetes Engine (GKE). GKE provides containerised application management, powered by Kubernetes, which includes load balancing, automatic horizontal scaling, and upgrades. A globally distributed fleet of GKE clusters ensures customers receive the same Loqate standard of service irrespective of where in the world their requests originate from. 

In addition to GKE, Loqate leverages many other Google Cloud resources to achieve greater levels of scale and reliability. A few examples:

  • GCS and Filestore are the primary sources of storage for our vast amounts of location data
  • Pub/Sub and MongoDB, in combination with an event-driven architecture, enable real-time streaming of usage events to provide accurate billing and reporting
  • Google CDN helps millions of end-users around the world load our low-overhead JavaScript tag in milliseconds
  • Cloud Monitoring, Logging, and Trace ensure we can observe our platform health in real time, track performance against our service level objectives (SLOs), and  

Working together 

Of course, Loqate isn’t just code running in Google Cloud; there are humans who work to make all the magic happen. In addition to building a secure and scalable product, Loqate also strives to build a generative work culture with high cross-functional collaboration and shared responsibility for quality, availability, reliability and security.