How does one add Redundancy/High Availability to AWS Lambda / Serverles

I’ve created a Shopify CarrierService using serverless v0.5. Essentially what that is something that calculates shipping costs for orders. I use this on multiple stores and it’s been great.

Problem is AWS Lambda occasionally dies for couple hours at a time. When this happens, my customers are not able to receive shipping costs, and thus can not finish their orders. When this happens I lose thousands of dollars and so do others using my service for their stores.

With that said and the larger question for people using serverless is how could one build in redundancy / high availability when using AWS Lambda.

Just figured I’d toss this over to the community to gather suggestions.

Depending on how fast the calculations need to be you could run it in several datacenters and then have eventually consistent results, e.g. calculate in 2 datacenters (or have fallback only) and compare the results of both. If they are fine everything is good, if one datacenter dies not a problem because it gets calculated twice. Depends on how expensive your calculations are and if calculating twice would be fine.