Please visit here regularly for new items, or subscribe to the RSS feed.
04 March 2013
Major New Version of Mediaflux Available Soon
Version 3.8 of Mediaflux, to be released soon, contains major storage orchestration enhancements as well as other improvements.
Take data to the compute or ... take compute to the data
The Mediaflux pledge for data to be easily found and delivered to the right place at the right time takes on additional significance.
Mediaflux now supports multiple storage pools, with multiple copies of data, policy driven distribution, affinities, cloud storage, prioritised I/O, and virtual storage tiers that can be dynamically added and removed.
Together, the new capabilities add up to unprecedented flexibility managing data locality, for both redundancy and affinity purposes.
Content Store for Amazon S3 Cloud Storage
A new Content Store type has been added to support asset content on Amazon S3 cloud storage. The store includes support for quota management to allow multiple stores per S3 account, as well as support for proactive high-water response.
Content Store Policies
Content Store Policies have been introduced to allow for the automated distribution of assets across multiple storage sources, whether local or remote. A policy is comprised of the set of potential content stores and a policy type. Policy types include balanced, round robin and waterfall.
These features can be used to balance data amongst a group of content stores, round robin data between stores, or automatically spill data to another store when a store fills, offering tremendous flexibility in the movement and location of data.
Content Store Affinities
The content for an asset can now be stored in more than one Content Store. This allows content to be duplicated in multiple Content Stores (which may be at different locations) for redundancy and affinity purposes.
Prioritised Store I/O
Prioritised Store I/O allows the order of I/O requests to be altered. It is very useful when multi-tier storage is in use and can be configured to prevent overflow of the disk cache, leading to poor I/O performance.
Content Store High Water Mark
Content Stores now support high water mark level service calls, where a specified service may be called when a high water mark is reached. The service can take any action required, whether remedial, notification, or some other action.
Virtual Namespace Hierarchies
A Virtual Namespace Hierarchy is an indexed collection of assets generated dynamically based on metadata.
Virtual Namespace Hierarchies are a dynamic form of namespace that are fast to query, due to indexing, and can be exposed by the Mediaflux NFS server or as an Arcitecta Asset File System. Virtual Namespace Hierarchies allow the same asset to appear in multiple namespaces, depending on its metadata.