Following #91 , it would be nice to add some caching. When using an include the file is fetched from S3 every time, so running lots of commands under treehugger means lots of network traffic to S3. Since we only support versioned S3 URL's they are guaranteed to not change, so we could cache them indefinitely in a 'treehugger' app dir to improve the launch time for 'treehugger exec'.
That said, treehugger does have to talk to KMS for every encrypted variable, so it's only saving one of several network hops. However KMS is a faster and less variable than S3.
Following #91 , it would be nice to add some caching. When using an
includethe file is fetched from S3 every time, so running lots of commands under treehugger means lots of network traffic to S3. Since we only support versioned S3 URL's they are guaranteed to not change, so we could cache them indefinitely in a 'treehugger' app dir to improve the launch time for 'treehugger exec'.That said, treehugger does have to talk to KMS for every encrypted variable, so it's only saving one of several network hops. However KMS is a faster and less variable than S3.