pulls s3 bucket usage per storage engine

s3-stats.py

import boto3
from datetime import timedelta, datetime


### Logic:
#  this script will list all aws buckets, and will retrieve each bucket size per storage type,
#  output will be displayed in mb, if output is 0Mb, the specific storage size is less than 1Mb.
#  output will only display used storage types
###

class Aws:
    def __init__(self, aws_region='us-east-1', profile='default'):
        self.region = aws_region
        self.session = boto3.Session(region_name=aws_region, profile_name=profile)

    def get_bucket_stats(self, bucket_name, bucket_location, storage_type):
        cw = self.session.client('cloudwatch', region_name=bucket_location)
        response = cw.get_metric_statistics(
            Namespace='AWS/S3', MetricName='BucketSizeBytes',
            StartTime=datetime.utcnow() - timedelta(days=2),
            EndTime=datetime.utcnow(), Period=86400,
            Statistics=['Average'], Unit='Bytes',
            Dimensions=[
                {'Name': 'BucketName', 'Value': bucket_name},
                {u'Name': 'StorageType', u'Value': storage_type}
            ])
        if response:
            try:
                ret = response['Datapoints'][0]['Average']
            except Exception as e:
                ret = 0
            return ret
        return 0

    def get_bucket_location(self, req_bucket_name):
        response = self.session.client('s3').get_bucket_location(Bucket=req_bucket_name)
        bucket_loc = 'us-east-1'
        if response['LocationConstraint'] is not None:
            bucket_loc = response['LocationConstraint']
        if response['LocationConstraint'] == 'EU':
            bucket_loc = 'eu-west-1'
        return bucket_loc


if __name__ == '__main__':
    # us-east-1 - for api calls
    # default - default profile in ~/.aws/config
    aws = Aws('us-east-1', 'default')
    buckets = aws.session.client('s3').list_buckets()
    if buckets:
        for bucket_list in buckets['Buckets']:
            bucket_name = bucket_list['Name']
            try:
                bucket_location = aws.get_bucket_location(bucket_name)
            except Exception as e:
                print(str(e))
            types = ['GlacierStorage', 'GlacierStorageOverhead', 'IntelligentTieringFAStorage',
                     'IntelligentTieringIAStorage', 'OneZoneIASizeOverhead', 'OneZoneIAStorage', 'ReducedRedundancyStorage',
                     'StandardIASizeOverhead', 'StandardIAStorage', 'StandardStorage']
            print(f'region: {bucket_location}, bucket name: {bucket_name}:')
            bucket_total_size = []
            for storage_type in types:
                bucket_size = aws.get_bucket_stats(bucket_name, bucket_location, storage_type)
                bucket_size_in_mb = round(bucket_size / 1024 / 1024)
                if bucket_size:
                    bucket_total_size.append(bucket_size_in_mb)
                    print(f'    Storage type: {storage_type}, size in Mb: {bucket_size_in_mb}')
            if bucket_total_size and len(bucket_total_size) > 1:
                bucket_total = sum(bucket_total_size)
                if bucket_total > 1048576:
                    bucket_total = round(bucket_total/1048576, 2)
                    print(f'Bucket total size in \033[1mTb\033[0m: {bucket_total}')
                elif bucket_total > 1024:
                    bucket_total = round(bucket_total/1024, 2)
                    print(f'Bucket total size in \033[1mGb\033[0m: {bucket_total}')
                else:
                    print(f'Bucket total size in Mb: {bucket_total}')
            print('\n')
    else:
        print('Couldn\'t retrieve bucket list')