Authored by Nils Haustein and Madhu Thorat
IBM Storage Scale version 5.2.1 provides a new S3 object storage service that builds upon the outstanding capabilities of IBM Storage Scale. IBM Storage Scale S3 Service provides high performant, and scalable S3 object access to data that is stored in IBM Storage Scale file systems. S3 buckets and objects are stored as directories and files in the file systems and can be tiered to tape.
Tiering to tape leveraging the IBM Storage Scale policy engine is flexible, fast and predictable. Policies allow an administrator to define the lifecycle of objects in a programmatic manner. Policies are executed by the policy engine.
With policies it is possible to tier objects from certain or all buckets to tape. Furthermore, objects from different buckets can be tiered according to different criteria. For example, objects in a bucket named work are tiered to tape when they were not accessed for 90 days. Objects in a bucket named archive are tiered to tape when they were not accessed for 10 days.
Likewise, tiering criteria can consider the size of the objects. Tiering small objects (< 10 MB) is not as space efficient than tiering larger objects (> 1 GB). It is also possible to tier objects in certain S3 storage classes, or objects with certain S3 metadata and tags.
Are you curious how IBM Storage Scale S3 object storage tiering to tape works? Take look at this comprehensive paper:
IBM Storage Scale S3 object storage tiering to tape
This paper gives a comprehensive introduction for S3 object storage tiering to tape with IBM Storage Scale S3 Service and tape management components such as IBM Storage Archive Enterprise Edition or IBM Storage Protect for Space Management. It guides you through the architecture and deployment of IBM Storage Scale S3 Service and tape management components. It explains how to create S3 accounts, buckets to write, read, update and delete objects through the S3 service.
This paper provides comprehensive guidance for managing S3 object storage tiering to tape by leveraging the IBM Storage Scale policy engine and the tape management component. Tiering means that files are migrated to tape while allowing transparent access to migrated files in the file system. When migrated files are accessed as objects by the S3 user, then the content of the files is recalled from tape to disk transparently.
Special emphasis is put on transparent recalls and how these can be prevented to avoid recall storms. Transparent recalls are caused by S3 GET requests for objects that are migrated to tape. Many simultaneous transparent recalls can cause recall storms and slow down the object storage system. For this reason, many transparent recalls should be prevented and replaced with tape optimized recalls.
Finally, this paper provides best practices for IBM Storage Scale S3 object storage tiering to tape.