The client needed a complex data migration that was going to be costly. We found an entirely different serverless approach that saved time and money.
Years ago a client came to us needing to migrate a serverless Python solution from Vercel Now V1 to Vercel Now V2. There was some machine learning data and a model that needed to properly load, and quickly into the migration we realized that V2 was a different architecture that had much more imposing space limitations due to AWS Lambda under the hood (this was years ago before more recent improvements). While the codebase could have potentially been rewritten or re-architected, it would have been somewhat costly.
The big win here was that we suggested to the client an entirely different serverless approach using Zappa’s powerful S3 pulling functionality to get around Python packaging size limits. We honestly weren't sure if it would work, but it seemed like the only straightforward way of keeping the system as is without having to change much of the code around. Fast forward, and 14 documentation steps later, we had a working deployment with Zappa, getting around Lambda’s limitations.
Even if a client wants something done in a particular way, we are more than happy to explore alternative solutions with clients, and those alternative solutions may save lots of money while providing an identical or better experience than what you were expecting.
