| Randy Kerns
Partner, the Evaluator Group
| Randy Kerns is a partner at the Evaluator Group and is responsible for storage area networks (SAN) and network-attached storage (NAS) analysis and education as well as company and product strategies. He has over twenty-eight years storage product development, including work for IBM, Fujitsu, Vice President of Engineering at the Array Technology subsidiary of Tandem Computers and Director of Engineering for Enterprise Disk at Storage Technology Corporation.
There have been several Fibre Channel switch announcements recently that have set new price points for the per-port cost that a customer pays. The per-port price is for a switch (or director) based on F-ports that are fabric ports that connect to other elements such as storage devices and servers via host bus adapters (this does not include Inter-Switch Link – ISL ports that are typically more costly). In general, the per-port cost is arrived at by dividing the switch price by the number of ports.
The prices are list prices and rarely does a customer pay list price. After discussing the pricing paid by many customers, it is clear that most expect to get about a 50% discount of some type. The discount any customer gets depends on the size of the purchase and how badly the vendor wants the business.
The price declines for switches are very welcomed by customers and are crucial to more broad Fibre Channel adoption. This follows the normal technology pattern where there is competition, as technology matures and volume increases, the price comes down. The continued downward price decline has clearly been the case in Fibre Channel switches. There is an area where this has not followed the pattern, however. That area is the price of Fibre Channel host bus adapters. The price of HBAs should parallel the price declines of the Fibre Channel switch ports. Actually, it should be on a steeper price reduction curve due to greater volume due to usage in non-switched environments. Why hasn't this been the case?
This is more than just an interesting question. It smacks of something that doesn't pass the smell test. If it defies logic, there has to be more to it. Most of us that have been in product development can give a fairly good cost estimate for products based on the components. Factor into that the support (including interoperability testing) and warranty costs along with cost of sales and administration as a percentage of product cost and we should have what the vendor can sell the product at and break even. After that, we have profit. Included in the profit should be a reasonable amount of money for the opportunity investment that was made. So, why do the HBAs still cost so much and why hasn't the price decline at the very least mirrored that of the per-port cost of switches?
Usually it is the customer that can put pressure on the vendor to meet price expectations. In the case of HBAs, the purchase is done by the company that puts a solution together to market to the end customer. This means that the end customer is usually seeing a bundled price that includes many other items such as switches/directors, software, cabling, servers, storage, etc. To let free enterprise work in this scenario, either the end customer has to break out the choice of HBA (and the individual pricing) to allow a competitive choice to be made where price is a determining factor or the solution provider has to play that competitive game more aggressively. I'm not sure what the correct answer is, I just know that there is an imbalance in pricing declines over time.
Most people understand that it is not in the vendor's best interest to see the price declines that come with volume and maturity. To avoid this or at least postpone it, vendors seek differentiation by several methods where that differentiation can command a price premium. Continually adding new differentiation continues the postponement. A good example here is the advanced functions that have been added to fabric switches. We're now seeing additional security and management functions that have been added by the different vendors. This creates a difference that can be marketed used as a reason to justify additional pricing or as a selection criteria element.
Unfortunately, in some switches these are implemented as special features that make them incompatible with other vendor switches. To fully interoperate, these advanced features may have to be disabled (through settings or a different code load). So, they are differentiable, command a premium, and potentially create a vendor advantage (lock-in if you consider it from the negative side). Tighter standards with fewer areas for "flexibility" would help here and vendors that promote that should be encouraged.
The economics of technology maturity and volume are winning out with Fibre Channel switches as it has become a very competitive battleground. Now if we can just get the HBAs to follow that same trend.