Google Cloud recently announced the availability of Autoclass for Cloud Storage, a bucket-level setting that automates the lifecycle management of objects based on their last access time. The new feature simplifies cost optimization for workloads with unpredictable and unknown access patterns.
Autoclass automatically transitions objects in a bucket based on each object's access pattern, supporting the standard, nearline, coldline, and archive storage classes. The new service manages all new objects starting in standard storage and transitioning them to progressively colder storage classes after 30, 90, and 365 days if they are not accessed. Sonit Tayal, product manager at Google, and Nishant Kohli, senior product manager at Google, explain:
Data placement can have a significant impact on cost as the storage price for the archive class is 6% of the storage price for the standard storage class. Having frequently accessed data in colder storage classes can lead to unexpected operations charges. Having data that is never accessed in standard storage can lead to large at-rest storage bills.
Source: https://cloud.google.com/blog/products/storage-data-transfer/optimize-your-cloud-storage-spend
In exchange for a management fee, the service helps achieve price predictability by removing retrieval surcharges associated with colder storage classes and it is recommended if the data in a bucket has a variety of access frequencies, or if the access patterns for the data are unknown or unpredictable. However, it is not recommended if the data fits into the use cases of a specific storage class, for example, backups. Tayal and Kohli add:
All transitions carried out by Autoclass can be observed using Cloud Monitoring dashboards. This provides easy observability and allows you to validate what Autoclass is doing for your data in an automated way.
The service can be monitored using the metrics autoclass/transition_operation_count, the number of storage class transitions automatically initiated, and autoclass/transitioned_bytes_count, the total number of bytes transitioned.
Google Cloud is not the only provider offering a service that automatically moves data across storage classes with S3 Intelligent-Tiering introduced by AWS in 2018. Forrest Brazeal, head of developer media at Google Cloud, tweets:
Yay Autoclass! This is similar to S3's Intelligent-Tiering feature - identical pricing, in fact. The biggest difference, I believe, is that Autoclass is a bucket-level setting whereas Intelligent-Tiering can be done at the level of individual objects/prefixes.
Buckets with Autoclass are charged a monthly management fee of 0.0025 USD for every 1000 objects stored, with no retrieval fees for accessing an object and no early deletion fees.
Currently, Autoclass must be enabled at bucket creation time using CLI, API, or Cloud Console. Support for existing buckets is expected later in 2023. Disabling the feature, all the objects in the bucket remain in their current storage class.