Allow HOURLY granularity

[iakov-aws]

Some customer have a big CUR. To improve performance we suggest a workaround to change time granularity to daily to reduce 24 times the volume of data.

TimeUnit: HOURLY

Need to update legacy and data collection:

• https://github.com/aws-samples/aws-cudos-framework-deployment/blob/main/cfn-templates/cur-aggregation.yaml
• https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-resource-cur-reportdefinition.html#cfn-cur-reportdefinition-timeunit

Make sure to:

• allow only allowed values
• have a default as HOURLY
• explain in the doc that HOURLY is recommended
• update 3 fields
• bonus points for the same for data-exports

Current workaround: Granularity can be changed added manually after deployment (and before backfill). Or (preferably) by changing the code before deployment.

- TimeUnit: HOURLY
+ TimeUnit: DAILY

[yprikhodko]

++ include explanation in the doc the change process from hourly to daily:

1. Delete current stack in payer account (Update of the stack won't work as TimeUnit is a Replacement parameter and replacement CloudFormation cannot update a stack when a custom-named resource requires replacing. Rename cid and update
2. Delete S3 bucket with current hourly CUR in payer account
3. Delete hourly data in S3 bucket in data collection account
4. Deploy https://github.com/aws-samples/aws-cudos-framework-deployment/blob/main/cfn-templates/cur-aggregation.yaml stack in payer account and choose Daily as parameter
5. Request backfill of the new CUR via support case

[iakov-aws]

So basically we need to remove and create the stacks in both source and destination accounts with full bucket removal. This is quite an update 😓
.

[yprikhodko]

No need to re-deploy stack in destination account, only data cleanup to delete/backup old hourly data