Sepaton Update Tackles Large-Enterprise Database Deduplication

Just days after Hewlett-Packard laid claim to the deduplication crown in a bid to dethrone current market leader EMC, Sepaton is staking out its claim in the space. The company announced this week DeltaStor DBeXstream, a database deduplication software offering that tackles multistreamed and multiplexed data.

"The amount of data continues to grow in our customer base, anywhere from 25% to 100% per year," says Linda Mentzer, VP of product management and marketing at Sepaton. Large enterprises, particularly Oracle shops, developed scripts to stream to multiple tape backups, she says--problems arise, however, if enterprises try to stream to disk, as data tends to get scrambled. "We're the only vendor who can do this today--dedupe multistream and multiplexed data."

In addition to DBeXstream, DeltaStor 6.1 includes Symantec-certified support for NetBackup OST AIR and Accelerator, as well as Optimized Synthetics and Granular Restore. It also features new hardware-accelerated replication, increased daily system throughput and transparent space reclamation across all backup applications. Sepaton S2100 Data Protection V6.1 software is available immediately; S2100 systems start at $115,500.

Mentzer says the new 10 Gbit Ethernet support means customers can do restores in smaller time windows or support more data. The other new features reduce the amount of data that needs to be backed up, saving both time and disk space.

Based on initial customer trials, performance increases by a factor of two and throughput by about 20%, she says. It's a software-only release, she adds, so the upcoming availability of storage and servers within the next six to 12 months will provide another dramatic performance boost, as well as new data protection functionality.

Deni Connor, principal at Storage Strategies Now, says the offering's database deduplication capabilities will set it apart. "The software also allows database administrators to set more granular backup--of sub-8K blocks--without backup and recovery penalty. These differentiations between Sepaton's DeltaStor and that of the hash-based deduplication vendors is important because it lets database admins backup and restore data as they can traditionally do it with tape-based systems."

Jason Buffington, a senior analyst at Enterprise Strategy Group, says that while the new announcements will appeal to Symantec customers who should be happy to see an additional storage option, he agrees that the bigger news is DeltaStor DBeXtreme. "If you ask a DBA how to best back up large data sets, they will tell you to 'turn ON multistreaming,' whereby the data is fed to the backup solution in parallel streams of data; this capability is easily achieved through Oracle, DB2 or Microsoft SQL Server and needed when databases are truly of enterprise size."

However, a backup administrator will likely to tell you to turn off multistreaming, he says. "Why? Multistreaming a database backup breaks most deduplication mechanisms."

Because most backup solutions store each stream's data as if it came from separate data sources, data in "stream one" is stored as data source one, while data from "stream two" is treated as a separate data source. "Unfortunately, even with most fragment-based deduplication models, the same data that was sent in stream one the first night and then in stream two the second will appear like separate data--and therefore will likely not be deduplicated," he explains.

Sepaton's database deduplication solution doesn't use hashing; instead, it analyzes the data after receipt within the federated storage pool, Buffington says. "Because of this approach, it can discern redundant data between streams and iterations, enabling it to deduplicate the redundant elements that many other solutions cannot."

Sepaton's deduplication technology is unique because it eliminates the tradeoffs between backup performance and deduplication efficiency, enabling significant performance enhancements while providing major reduction in storage capacity for Oracle, SQL and DB2 databases, says Ashar Baig, a senior analyst and consultant at Taneja Group. "With Sepaton's DeltaStor DBeXstream's byte-level differential dedupe, their customers don't have to choose between multistreaming, multiplexing and capacity reduction through higher dedupe. Sepaton's customers can set data reduction ratios and storage utilization by client and backup job."

That this is a software release is good news for current customers, as they won't need to make a forklift upgrade, says Howard Marks, founder and chief scientist at DeepStorage and a Network Computing contributor. "Deducing appliances, whether NAS like DD or VTL like Sepaton, are Xeon servers running dedupe code, though some like Sepaton have compression cards, too. So the hardware upgrade will be to Romley."

Buffington added that it's reasonable to presume that Sepaton's 7.0 release would likely come with not only updated hardware, but also a generational update in hardware and versioning from Sepaton's OEM partners.