The most recent scaling challenge was onboarding a major client at Zapnito where infrastructure demands grew exponentially. While our initially response was to increase our infrastructure, we realised that we could make significant improvements internally.
Our email startegy also needed reworking, as we now send over 1.2 million user-specific emails every monthly. On our accountancy platform, end of year financial deadlines were particuarly high demand tasks which required making third-party APIs resilient with retries and failure handling.
To address these problems, the teams used several strategies:
We found that a few people had a particular interest in performance, so we leaned on their knowledge and ability to improve our platform.
Untlimately, there are many things to consider for scaling, so it depends on the problem being solved for and our target goals — maybe it’s okay if a nightly job that runs at 3am takes a little bit longer, if we’re working on a user visible problem that increases load times to multiple seconds.