Here's a summary of what you should do if you use AWS Lambda (Python). Including those that have not been tried yet. Updated from time to time.
-(First of all) I tried to summarize what to do after opening an AWS account https://dev.classmethod.jp/articles/after_get_aws_account/
--Applying the Jeffy framework Implementation of common processing for tracing and logging can be simplified. https://dev.classmethod.jp/cloud/app-fw-for-lamdba-jeffy-released/
--Apply AWS Lambda Destinations The following processing can be executed according to the execution result of Lambda. https://qiita.com/kojiisd/items/efcb2ac3d5cc176534ba
--Apply AWS Lambda Layers You can standardize frequently used libraries. https://qiita.com/t_okkan/items/394a15577bd1aad46ec3
--Parallel processing Parallel processing heavy processing. https://qiita.com/kenmaro/items/69fdd84e18e793a21790 https://aws.amazon.com/jp/blogs/news/parallelizing-across-multiple-cpu-gpus-to-speed-up-deep-learning-inference-at-the-edge/
--Apply AWS Lambda Power Tuning Optimize memory allocation. https://dev.classmethod.jp/cloud/aws-lambda-power-tuning/
--AWS Lambda cost savings Introduce a profile tool. https://dev.classmethod.jp/cloud/aws/lamba-profile-with-qcachegrind/
--Pip update (added to buildspec.yml)
#View a list of packages with updates with pip
pip list -o
#Batch update with command
pip list -o | tail -n +3 | awk '{ print $1 }' | xargs pip install -U
Recommended Posts