Original URL: http://www.theregister.co.uk/2010/10/20/arkeia_nwbu_v9/

Arkeia smashes through the blocks

'Progressive' dedupe technology makes 'faster' mousetrap

By Chris Mellor

Posted in Storage, 20th October 2010 07:00 GMT

Arkeia is adding deduplication to its entire line of backup software, appliances and virtual appliances with Network Backup 9.0.

The deduplication is in-line, as source de-dupe must be in order to gain the advantage of reduced network bandwidth when transmitting the deduplicated backup files to a central store. Arkeia's deduplication technology is based on technology from Kadena, which it bought in 2009.

Arkeia calls it progressive deduplication, saying it doesn't rely on fixed or variable block boundaries, and claims it is faster than either of these methods, eliminating any need to scan for block boundaries.

Arkeia says: "All files previously encountered by Arkeia are deduplicated at fixed-block speeds. New data is surveyed with a sliding window.  [Its] algorithm determines if data under the window is a probable match to blocks in the known-block-pool. Probable matches are scrutinised with a heavy-weight hash algorithm. Because over 99 per cent of probable matches prove to be exact matches, [such] progressive matching is extremely efficient."

Arkeia reckons this is better than either fixed or variable block deduplication. It said: "Progressive deduplication evaluates all possible block boundaries, guaranteeing the best possible deduplication. Precise control over block boundaries allows blocks to be optimally sized for each file type. Each type of data, such as executable files, text files, and database records, will be deduplicated with the block size that achieves the maximum compression rate. Arkeia has analysed hundreds of file types, produced by hundreds of popular applications in the enterprise, to determine each one’s optimal block size."

So there you have it: Arkeia's dedupe mousetrap is faster than other dedupe software mousetraps running on similar hardware and it produces better dedupe ratios as well.

An Arkeia spokesperson says it's great for backing up virtual servers: "We tested dedupe of virtualised disks (e.g. VMDK files) with different block sizes, training our software to dedupe virtualised disks with the optimal block size for maximum compression. … We shipped a new backup agent for VMware’s vSphere, using vStorage APIs, back in June and now support vSphere 4.1 including CBT and vCenter. Our dedupe is intended to be used in conjunction with this and Arkeia’s other specialised agents."

Arkeia's Backup Agent for vStorage uses VMware’s changed block tracking (CBT) to backup, restore, or replicate virtual machines. For Microsoft’s virtual environments and applications there is Arkeia’s VSS-based Backup Agent for Hyper-V, and Network Backup 9.0 also does agentless backups and restores of Citrix XenServer environments.

All Arkeia appliances can be upgraded to version 9.0 firmware, meaning current appliance users can get source-side deduplication.  No hardware upgrades are necessary because source-side deduplication leverages the processors at the client computer to compress data before it travels over the network. You might notice your CPU cycles when it is used in his way though.

Network Backup v9.0 includes encryption, an improved web user interface and wider platform support. It will be generally available in the first quarter of 2011 for Linux and Windows platforms.  Support for AIX, BSD, HP-UX, Macintosh, Netware, and Solaris will follow in 2011. The Deduplication Option is priced at $2,000 per media server and Deduplication Option licences will be bundled at no charge with all new backup servers, software or appliance, purchased by 31 December, 2010. ®