Big databases need big storage.
To that end, Oracle this week announced the Exadata Storage Expansion Rack which has the potential to deliver as much as 10 Petabytes of compressed storage for the Exadata Database machine for clustered deployments.
Exadata is Oracle's all-in-one database box that includes compute, networking and storage capacity. The Storage Expansion Rack provides additional storage to Exadata via redundant 40Gb/sec Infiniband connectivity
Tim Shetler, VP of Exadata product management at Oracle told InternetNews.com that the storage on an in-service Exadata machine is being used for current transactional data or the latest load from a data warehouse.
"Over time that data becomes less hot or current but you might still have a reason to keep that data around," Shetler said. "You could put this off to tape but that would be an additional step."
Shetler noted that the Exadata Storage Expansion is for data is that is less current but is still needed online with the performance of an Exadata machine.
In terms of capacity on a full rack of the Exadata Storage Expansion has capacity for 432 TB of raw storage data. Exadata has a feature called Hybrid Columnar Compression which provides roughly 10x compression so 432 TB turns into 3 Petabytes of compressed storage.
Shetler explained that up to eight database machine and storage expansion racks can be connected together. With one database machine and seven storage expansion racks, Exadata can support 10 Petabytes of compressed data.
Shetler noted that prior to the new Exadata Storage Expansion appliance an customer could expand storage by buying another Exadata box and not license the compute nodes. The other way to do is that Oracle sold separate storage servers on their own, but users would have had to add their own Infiniband cables and integrate it themselves. The Storage Expansion rack in contrast is a pre-packaged and pre-integrated system.
In terms of moving data around, Shetler said that Oracle Database makes it easy to move data around as part of an information lifecycle management strategy. He explained that Oracle recommends that customers create partitions that vary based on time for date ranges.
"So they would create a partition for current data and then automatically for data in the table that have dates in them, the data would automatically be put in the right partition," Shetler said. "So older data could automatically move to the expansion rack over time."