Italic is an e-commerce marketplace that connects high-end manufacturers directly to customers, which allows you to shop straight from the same factories as high-end brands but pay up to 75% less. Founded in 2018 with the idea that customers shouldn't have to pay a premium just for a label, Italic seeks to create a world where everyone can afford to live well.
With the ongoing high demand and fast growth of their offering came serious scaling problems. During 2020's Black Friday, the website crashed every few hours as their backend was overwhelmed by the traffic spike.
Let's dive into Italic's architecture to understand how they got ready for Black Friday 2021.
The original version of the marketplace was heavily reliant on third-party services. They used Shopify for their product and collection management portal, while they utilized Yotpo to collect product and customer reviews. While using those services helped them get their product off the ground quickly, they ran into many limitations for their specific use case — thus, they decided to own these systems in-house.
Over the course of 2021, they built their own product management, collection management, and product and customer reviews services from scratch. All of them use a modern stack based on Go, Node, and TypeScript, and the communication between client and server happens via a central GraphQL API — which is perfect for the kind of data e-commerce websites deal with.
In order to prepare for 2021's Black Friday, the team knew that they needed to change something. Most of their highly requested data, like products, is the same for every user — an ideal situation for caching.
They looked into GraphCDN's GraphQL edge cache, which seemed like a perfect fit. They got setup within less than a day, and almost immediately after deploying to production, Italic's most important and highly requested queries had an 80%+ cache hit rate, with some even reaching 98%!
That, in turn, meant their database pressure was reduced by about two orders of magnitude, their overall server load was reduced by 61%, and their page load times were reduced by over 1 second.
Not only did they survive 2021's Black Friday traffic spike without any downtime, but their infrastructure had so much room to breathe that they're now prepared for any upcoming traffic spike thanks to the caching.