Hey Folks, I am looking for some advice on memory recommendation on lambda cold starts for Python specifically. It looks to me like some of our cold starts are taking more than 1 second while using 512 memory. I want to bump it up, but not sure if I should go to 1024 or much greater.
With C# and Java we had to go up to like 3k to get cold start times way down, but then it actually ran the stuff WAY faster, so it saved us money long term. I don’t think it’s the same case for Python. It may mostly be a moot point with my healtchecks keeping things warm.
Could you guys suggest me specific memory recommendations to python lambda function keep warm. Great Thanks for your precious time!
You could use provisioned concurrency
Often you need to look at your dependencies and see what is taking so long to load. Is it lambda or is it some dependency that is taking the 1 second. Moving up to 1024 will definitely help as well.
Have you seen https://github.com/alexcasalboni/aws-lambda-power-tuning ?
Great Thanks , and for suggesting so many solutions.
I am trying to understand above solutions one by one and put efforts to fulfil our use-case.
BTW!, I have seen it once but not tried. Now i will try and take necessary action
1 full cpu is also ~1700, not 1024
Going beyond that usually doesnt help with python/ruby/etc
Thanks for your suggestion
My team thinks for now we go with 1024 MB in place of 512 MB
We should try power tuning in the future https://github.com/alexcasalboni/aws-lambda-power-tuning
