3PAR, a pioneer of thin provisioning technology, has embraced the market for storage for server virtualisation and is orienting itself towards the emerging desktop virtualisation space. The company recently launched its T-Series InServ line of storage arrays, which feature silicon-based thin provisioning capabilities that aim to improve the performance of the virtual allocation technology.
Since being founded in 1999, 3PAR has only this year made a profit – at the start of what might be a severe economic recession. So, while the California-based vendor is clearly in the vanguard of storage technology, it seems reasonable to wonder what an uncertain economic environment might do to a company with so little to show in terms of profit.
SearchStorageUK: 3PAR has only just made a profit for the first time. It's a bad time – economy-wise – to be only just making a profit. Why will 3Par survive a major recession, periods when industries usually see major bouts of consolidation?
Scott: We believe there is a unique opportunity for 3PAR to become the third major storage company after EMC and NetApp because large independent IT companies emerge when changes occur in the architectural environment. EMC arose on the back of mainframe computing, NetApp rode the wave of client/server architectures. 3PAR has the same opportunity with utility computing. It's a real opportunity and we are taking large amounts of market share and have an annual growth rate of 80%.
Scott:Utility computing is enabling a major transformation in enterprise IT delivery, bringing a cloud computing-type delivery model that enables both software- and hardware-as-a-service from a central location.
In the next 15 to 20 years we will see enterprise IT change like industry's use of power did in the early part of the 20th century – from each factory having its own steam engine in the 1900s to everyone buying electricity as a service from the 1930s. What is needed to create a utility storage environment is to generate a shared virtual environment for flexible workload consolidation.
SearchStorageUK: How does one accomplish that?
Scott: Three capabilities that traditional storage arrays lack are very important here. First, there is a need for agility that flows from the challenge of not being able to predict service requirements and the need to offer multiple levels of service. The core architectures of EMC, HDS and NetApp are not designed to give that agility and are preconfigured for a known deployment, such as an SAP implementation.
Secondly you also need tremendous efficiency. You win or lose as a service provider on the basis of cost-per-transaction at a given service level. Service providers look at all elements of cost and the problem with traditional storage arrays is that they suffer from poor utilisation.
The final requirement is resilience with security. Traditional high end arrays deliver good availability but need protection from other users and the possibility of disruption from the "accidental" loss of a LUN. To create a logical virtual private array, you need a virtualised OS, giving something like the chunklets, regions and pages we have that provide individual, local private arrays instead of the hardware partitioning that gives dedicated capacity individual customers.
SearchStorageUK: How do you feel when you see your competitors incorporating thin provisioning into their storage products?
Scott: In the last 18 months EMC, HDS etc. have incorporated thin provisioning into their products. But there are limits to how far the big storage vendors can do that with our technology. The problem for these companies is analogous to the problem of obesity. Likewise EMC, HDS and NetApp can't help being fat. Their architecture was never built with the granularity needed for this. To them, a disk is a disk.
With the 3PAR InForm operating system, we break disks down into 256 MB chunklets, 32 MB regions and 16 KB pages. EMC and HDS architectures naturally allocate much more, 42 MB in HDS's case.