您如何在无服务器架构中平衡nosql存储和缓存？

Question

I'm writing a go-based app for serverless environment (aws). Fairly new to writing app-level code, and I have an infrastructure background.

I understand cache isn't persistent, but do I really need to concern myself with persistence in a true serverless environment? From a requirements standpoint, simple dynamodb tables can accommodate my storage needs, but caching would give me better performance, right? And I need performance, which is a big reason why I'm using go for the app.

Answer 1

You could go for AWS ElastiCache (redis or memcache) but that isn't really serverless so may not be an option.

AWS Lambda allows you to write in /tmp directory and you could have something cached in memory. I do NOT recommend that option though. Lambdas don't always run in the same instance so you won't have any guarantee at all that next time the Lambda is invoked you'll have something on /tmp directory.

Take a look at Amazon DynamoDb Accelerator (DAX) that provides a highly available cache in front of DynamoDB.