OpenStack Object Storage (Swift) today released new capabilities called Storage Policies, the most significant feature update since the project’s inception. Storage policies allow enterprise deployers and application developers more control over how they want to store, replicate and access data across different back-ends and geographical regions, as well as the ability for public cloud hosting providers to offer several tiers of storage services.
Requested by enterprise users and cloud hosting providers, storage policies allow application developers to choose the storage hardware on which their data resides. Additional capabilities include specifying how data is stored across storage clusters and data centers in different geographical regions.
Users can also control replication policies that can improve the economy of different types of storage without compromising availability. Finally, storage policies add modularity and extensibility to the object storage system, allowing for new capabilities such as Erasure Coding expected later this year.
With storage policies, users can match their storage configuration to their application use case. For example, users might prefer spinning media for one application and SSD for another. Some applications might require triple data replication, while others need only 2x replication. And, some applications might need to specify use of storage clusters in a particular geographic location to meet data sovereignty and related regulatory requirements.
Storage policies have been in development for a year, and are the product of broad collaboration across the OpenStack community. More than three-dozen developers at 15 companies contributed to the version of Swift integrated into the current Icehouse release cycle of OpenStack. There are also several independent developers who contributed. Lead contributing companies (by lines of code) include SwiftStack, Intel, Red Hat / eNovance, IBM, Rackspace, HP, independent developers, EasyStack, SoftLayer and Dell.