Five network-attached storage (NAS) vendors looking to compete with NetApp's data deduplication for primary data have partnered with Ocarina Networks to integrate their products with Ocarina's ECOsystem storage-reduction appliances.
BlueArc Corp., EMC Corp., Hewlett-Packard (HP) Co., Hitachi Data Systems and Isilon Systems Inc. have all signed on for different levels of integration with startup Ocarina.
Ocarina Optimizer for BlueArc keeps the Ocarina ECO device on separate hardware, preserves file links in its original file system and compresses the data through Ocarina while moving it to lower tiers of storage. Hitachi Data Systems also sells the Ocarina Optimizer for BlueArc with BlueArc NAS systems that Hitachi OEMs.
HP's PolyServe and Isilon Systems have ported the software directly onto their clustered NAS heads to deduplicate data post-process, after it has been written to disk.
Carter George, Ocarina Networks' vice president of products, said vendors with clustered systems that split processing among multiple nodes have ported the software into those systems, while "more traditional ones want a co-processor in one of our appliances." He added, "There's some overhead for compression, and it will take up some cycles on the storage heads. If you let it, it will consume an eight-core server."
George said users can set their desired level of data reduction and performance in Ocarina Networks' software, which will be consistent across the different hardware configurations. The software will extract files from their original formats in all cases, decompressing any pre-compressed file formats and separating them into their component objects. From there, the software calculates a hash value for each object – a process known as "fingerprinting."
Users will have the option of using a "fast-path" process based on standard hash-based, block-level, post-process dedupe on objects. Or they can choose to apply standard "light" compression algorithms, as well as more heavyweight specialized compressors developed by Ocarina.
Until now, Ocarina Networks' appliances have applied "pretty heavy-duty compression for unique objects" that have kept it focused on archiving for media and entertainment customers dealing with large files, according to the firm's George. He said the vendor partners are all looking to level the playing field for primary storage data reduction with NetApp, which has had the feature available in its FAS arrays and V-Series gateways for approximately two years.
Users still kicking tires on primary dedupe
When the economic downturn began last fall, primary storage data deduplication became a coveted feature among end users, but few in the enterprise market are willing to be early adopters. The question remains whether organizations will allow a relatively new vendor to modify their primary data.
It depends on how hard users are squeezed by shrinking budgets while data grows during the recession, maintains Noemi Greyzdorf, research manager, storage software at Framingham, Mass.-based IDC. "There are those who really need footprint reduction, and those who would consider it if it was included in the price but don't have the pressure to really look for it," she wrote in an email to SearchStorage.com. "Those who are running out of space or need to extend existing resources' life, would look at these solutions as an option."
Another option for primary storage reduction is Storwize Inc., which also claims fast performance for primary storage data reduction. Ocarina Networks offers block-level dedupe post-process, while Storwize does in-band compression.
"I think the two products are not mutually exclusive and address different areas of the on-line storage market," wrote Greyzdorf. "Dedupe requires the data to be relatively static and do so as a post process because of the performance impact. Storwize compresses data inline and can do so for active data, adding more performance in some cases.
"I think that there is still a lack of good understanding of these capacity optimization technologies and how they work and where they work and whether they work together. A lot more education has to occur. I think we will also see the same vendors adopting the Storwize approach. The term dedupe has blinded many," she noted.
"It's all psychological," said Ocarina Networks' George of the post-process dedupe that leaves a "clean" version of the file to snapshot and back up. The Ocarina appliance could also sit in the data path, but storage administrators are more comfortable with it out of band, he said.