There is one crucial piece of data I was unable to find in the article, what memory size was Lambda set at?
For those unfamiliar with Serverless, you can choose how much RAM each function gets. Higher RAM costs more but also gives you a faster CPU. His Lambda functions are, on average, being billed for 500ms (according to one graphic).
If he is running with the base RAM allocation bumps it up he might see significantly faster performance and potentially billed for less.
Tuning lambdas and selecting the most optimal RAM (CPU) allocation is automatic with the help of tools provided by the community.
tl;dr Lambda is billed in 100ms increments and has configurable RAM. Higher RAM Lambda is more expensive but faster CPU. This can reduce your bill.
For those unfamiliar with Serverless, you can choose how much RAM each function gets. Higher RAM costs more but also gives you a faster CPU. His Lambda functions are, on average, being billed for 500ms (according to one graphic).
If he is running with the base RAM allocation bumps it up he might see significantly faster performance and potentially billed for less.
Tuning lambdas and selecting the most optimal RAM (CPU) allocation is automatic with the help of tools provided by the community.
tl;dr Lambda is billed in 100ms increments and has configurable RAM. Higher RAM Lambda is more expensive but faster CPU. This can reduce your bill.