Another thing that differentiates HP is its target deduplication architecture--that is, scale-out. She says it can grow from a single 48-Tbyte two-node couplet to up to 768 Tbytes of storage capacity based on a massive single namespace. "This allows for seamless scalability, enabling nodes to be added to the configuration without downtime. This is in marked contrast to first-phase scale-up deduplication appliances [what HP refers to as Deduplication 1.0] that, after hitting the maximum capacity threshold, require a disruptive forklift upgrade or additional stand-alone appliances to expand—introducing inefficiency, downtime and management overhead issues. This architecture also lends itself to high availability, which HP is delivering. Node failure is eliminated by pairing nodes within a couplet, so the surviving node can take over if its companion node fails."
Lastly, HP did some things differently in its deduplication method, says Whitehouse--specifically, smaller variable-sized chunks used for comparisons and a sparse index to deliver higher redundancy matching using a small, memory-resident index. This approach allows for a high reduction ratio without performance trade-offs. "More importantly, HP promises faster restore speeds. StoreOnce avoids a high degree of fragmentation by not replacing small amounts of duplicate data with pointers to faraway places with no other related data. Data is also defragmented after deduplication. The result is that restoring data takes less time because reconstituting it does not require many slow random seeks."
Whitehouse says the data dedupe market is in flux. The first wave of solutions were mainly hardware-based, but now deduplication is available in backup hardware and software solutions, so there is more choice. "HP can satisfy requirements for both hardware and software in a single solution. Additionally, they’ve leap-frogged competitors with backup performance [up to 28 Tbytes per hour), restore performance [28 Tbytes per hour) and price (HP claims its solution costs 20% less than rivals'). HP has a good footprint in the market. Their offering should get people’s attention and be disruptive."
Wikibon founder and consulting analyst David Vellante says HP's Information Optimization will have a heavy dose of Autonomy (the infrastructure software developer acquired a month ago for just over $10 billion). "Deduplication 2.0 is a big part of this for several reasons. Including deduplication naturally saves on space and fits into an optimization play. StoreOnce is IP developed by HP Labs, and HP Labs needs more commercial wins. The HP logo was "Invent," and HP needs to get back its invention mojo. StoreOnce is a start; HP was later to the market with its own data deduplication, so this is a big deal for HP customers in that they now have an HP home-grown solution, versus an OEM option or a non-HP option. This allows HP customers to wrap more efficient backup into a larger HP converged infrastructure deal."
The other thing is the vision of StoreOnce--that is, that you can deduplicate everywhere (primary, secondary and backup) without having to re-hydrate data, says Vellante. "This is a futures statement, but on paper HP could deliver on that vision and if it does it could have a big impact."
From a competitive perspective, Vellante says, this is about the emerging storage wars amongst the big whales (HP, EMC, IBM and Dell). "EMC is the dominant player in the deduplication market, with Avamar and Data Domain. It has about two-thirds of the market, and HP is intent on getting its piece of the pie. Just by having a solution under the HP brand, HP will gain share.”
See more on this topic by subscribing to Network Computing Pro Reports Research: Deduplication Grows Up (free, registration required).
