Original URL: http://www.theregister.co.uk/2012/06/27/datagravity/

Data-furtling execs look to establish cohabiting silos

It's all about contextual relationships

By Chris Mellor

Posted in Storage, 27th June 2012 09:43 GMT

EqualLogic alumni are at it again; running a storage startup to virtually tier data in different silos and extract information simply without complex server analytics software.

Nashua, New Hampshire-based DataGravity is led by Paula Long who co-founded EqualLogic. She and her co-founder John Joseph started up Data Gravity to turn data into information. Joseph was VP Marketing and product management at EqualLogic. Dell bought the company for $1.4 billion in 2008. Joseph served at Dell as VP storage solutions marketing for three years and then quit. Long served as Dell's VP of storage but quit earlier, in 2010. She joined a robotics startup in Boston but left.

Now they are together again, with $12 million initial funding. Long is DataGravity's CEO and Joseph its president. They see several storage silos in existence today; the existing block database data, unstructured data in filers, newer Big Data silos for business intelligence and emerging cloud-based silos for mobile and personal data and for off-site business recovery.

Getting information from the terabytes of data in these silos can need data extraction and processing by complex, server-based analytics software. DataGravity's idea is to form virtual storage tiers with silos co-habiting and being simpler to manage. Secondly it will provide a means to harvest information from this data without the complex, server-based applications needed today.

The company is recruiting for its operating system team and its distributed protocol team. A responsibility of the OS team is to "Develop enhancements, performance optimisations and patches for the Operating System Kernel, including networking, virtual memory, file system, or storage stack in the kernel."

The company reckons "There will be multiple distributed protocols required to complete the DataGravity vision. … In some cases an existing protocol may be the right choice, the Distributed Protocol will be responsible for enhancing these protocols to work with the DataGravity Solution." A responsibility will be to "Design and develop distributed protocols for the DataGravity product. These product will include high speed, low latency data transfer protocols, data access protocols and system management protocols."

A Boston.com report quotes Joseph: "People have been storing bits on disk drives for a long time with very little contextual information about them. We are thinking about the contextual relationship between information, and the solution a person might be looking for."

This suggests much metadata will be stored with the base data, to enable clever filtering and the provision of analytics functions in the DataGravity storage controller. The report then quotes Long; "We'd love not to have to build hardware, but we haven't ruled out creating an accelerator to do analytics on massive amounts of data. We're doing performance studies now to look at that."

Shorthand; think business analytics functionality in a storage array controller unifying data storage across various silos and providing masses of metadata to make the analytics functions simpler and faster. That implies a lot of indexing will be done to generate the metadata. ®