Storage software vendor Tarmin Inc. this week released a reworked version of its GridBank software with enhanced random access of objects, greater platform and application programming interface support, and performance profiling for storage tiers.
Tarmin refers to its GridBank 4.0 Data Management Platform as data-defined storage that is media-independent and allows customers to aggregate file systems into a global namespace.
Along with the global namespace to virtualize file systems, GridBank includes a distributed metadata repository and security and identity management. Joseph Lynn, Tarmin's vice president of marketing, said GridBank can access data from various hardware systems.
"Storage is separated into a lot of siloes, and this causes problems for organizations because they can't access data from a central tool," he said. "We ingest data from primary storage and then store it in a distributed object store."
GridBank previously supported Windows and client-based Linux, and now supports Linux on servers. It has been upgraded with metadata cache object storage identifiers that are stored in RAM within the storage nodes for better query performance.
"We can put metadata cache on high-performance nodes," Lynn said. "It reduces latency associated with typical object storage look-ups."
Tarmin also added a performance profiling migration function for storage pools and tiers, and it supports enterprise Linear Tape File System. It now also supports Amazon Simple Storage Service for cloud applications, Cloud Data Management Interface and OpenStack, while the GridBank proprietary sync-and-share application programming interface (API) is open to supporting third-party clients. A new API module has been built for social media data so that information can be ingested in the GridBank data repository. It also now supports offline email and PST Management, as well as Outlook Web App, a webmail service for Microsoft Exchange Servers.
Tarmin also added a post-process capability for tagging data, allowing GridBank to set data policies after ingesting data. Previously, customers had to set policies while data was ingested.
Tarmin has been around since 2007 and launched GridBank in 2009 as an archiving and e-discovery play, without a great deal of success. This year Tarmin moved its headquarters from the U.K. to Boston and signed a partnership with IBM to run GridBank on IBM's storage and System X servers.
Although there are plenty of object storage products available, Ashish Nadkarni, research director for storage systems at IDC, said Tarmin's data-centric strategy is a different way of approaching storage.
"The approaches taken by the majority of the storage industry has been to start with building a storage platform, and then force the data to conform to the characteristics of that platform," he said. "In a data-centric approach, the initiator and target roles are swapped. Rather than starting with storage and then defining how your data needs to fit, you start with data and define what your storage needs to look like based on that data."