DataDirect Networks today launched a higher-density version of its Web Object Scaler object storage appliance, as well as a software upgrade, adding integrated metadata
DataDirect Networks (DDN) also added the ability to replicate data from Web Object Scaler (WOS) to its GridScaler parallel file storage system.
The WOS7000 scales to 32 clusters, each holding up to 30 PB of disk capacity. Each 4U appliance holds 60 drives -- up to 2.64 petabytes (PB) -- and 11 drives fit into a rack. The appliances support 10 Gigabit Ethernet (GbE), 40 GbE and InfiniBand networking.
Tom Leyden, DDN director of product marketing, said a DDN WOS node can hold four billion objects and a cluster can hold one trillion objects across 256 object storage servers.
DDN also integrated a new search engine into its WOS core object storage software. The new version allows searches based on metadata. IT administrators define the key values for each object via a policy.
"It's a parallel back-end metadata search," Leyden said. "You can query one or more indexed fields of user-specified metadata on petabytes of objects. With this version of search, the end user can index up to five fields of user metadata. Boolean searches run on one or more of the indexed fields and return lists of object identifiers matching search criteria."
Leyden said parallel back-end searches are faster than local metadata searches and have no scalability limits.
"It's where big data analytics meets object storage," he said.
The DDN WOS platform also now integrates with GridScaler through a WOS Bridge module. The Bridge lets customers archive file storage along with objects on a WOS appliance. GridScaler data can be accessed through WOS or migrated back to its original platform.
More on the DDS WOS platform
DDN upgrades WOS 7000 object storage platform
DDN among best private storage vendors
Terri McClure, senior analyst at Milford, Mass.-based Enterprise Strategy Group, said the greater WOS scalability should appeal to service providers.
"Web service providers especially will love getting that much capacity in a namespace," she said. "The search makes the data useful by understanding what you have and why. That is becoming increasingly important. The data protection methodology they use is intriguing; the combination of local erasure coding with replication means good local performance."
"Object storage is still, in many ways, in early days, but for some use cases it will be a necessity and it will need to scale to these levels," McClure said.
DDN's Leyden said he expects object adoption to pick up soon.
"80% to 90% of growth we are seeing is unstructured data," she said. "And there are many use cases for object storage."