Configure Storage Classes for Buckets

[tjholm]

It is a very common requirement to configure storage classes on buckets based on the performance and access requirements of objects stored in that bucket.

Ultimately storage class is determined at the object/file level, but being able to set a default for a bucket when creating new objects makes sense.

[tjholm]

There are few options for this we could include this in our code/apis, or include as additional configuration.

Using code:

import { bucket } from '@nitric/sdk';

const files = bucket('files', {
  class: 'standard'
});

// This could also be configured at an environment level (e.g. dev/staging/prod) using ENV variables)

const STORAGE_CLASS = process.env.DEFAULT_STORAGE_CLASS || "standard";

const files = bucket('files', {
  class: STORAGE_CLASS
});

// OR could support provider specific configuration if necessary (rather than with an abstract mapping)

const files = bucket('files', {
  aws: {
    storageClass: 'INTELLIGENT_TIERING'
  },
  gcp: {
    storageClass: 'NEARLINE'
  },
});

Using config (nitric stack file):

buckets:
   default:
      class: standard
   my-bucket-name: 
      class: cold

Extra things to consider:

• Should class implementation be abstracted or be configured for specific providers? (or both)

[tjholm]

With new provider configuration available, this should be relatively straightforward to implement for named buckets.