Posted by: Peter Burrows on December 31, 2005
In a recent article on Network Appliance CEO Dan Warmenhoven, I talked about Warmenhoven’s “state-of-the-shelf” strategy for product development. This is the idea that NetApp relies on hardware components made by others, and combines it with its own software to differentiate the products. A reader wanted more information on this, so I pinged Dan, who was kind enough to blackberry a clarification while on vacation. Here’s what he said:
The “state of the shelf” strategy has to do with how the systems are designed and what components we use. We have never designed a custom semiconductor or ASIC. Most other storage manufacturers have several ASIC’s in each system. EMC’ systems have quite a few.
Most of our hardware sub-assemblies are also sourced from other manufacturers. Our disk shelves and packaging are sourced at Xyratex, a company based in the UK. The system mother boards and chassis for our FAS900 series were co-designed and are manufactured by NEC who sells tham as Wintel servers. The I/O cards such as the Ethernet NICs and FibreChannel HBA’s are standard cards from suppliers such as Q-Logic.
However, the system we take to market is uniquely NetApp. Part of our business practice is that all elements of the system must be purchased from NetApp by the customer. We do not allow or support the introduction into our systems of components or disk drives purchased from sources other than NetApp. There are several reasons for this policy (including revenue), but the primary reason for this policy is quality and support. Our systems are hihgly tuned high performance systems that require careful quality control of the components to function properly and reliably. Every time we have allowed customers or partners to plug in components or disk drives from other sources we have significant reliability issues. And when an issue occurs we are not able to provide support because we do not know what is in the system.
So the components and sub-assemblies are “state of the shelf”, but the way we put them together into a system is part of our value add.
In a second email, Dan added that:
In no way does the usage of “off-the-shelf parts” imply that the resultant system product is a “commodity white box solution”, and never have we characterized our products as a “commodity solution.” The notion that using commodity components, i.e., “parts” yields a “commodity solution” is a total non-sequitor. Our systems are far from commodity. They are carefully engineered for high reliability, high performance and high serviceability. There is nothing “commodity” about the system, only the components that comprise it.
Thanks for the question, FactCheck, and thanks for the response, Dan.