Cisco Blogs

On Merchant Silicon and Mowing My Yard

March 6, 2008 - 8 Comments

Recently there was a quick write-up by one of my favorite competitors in the switching market arguing against my assertion that ‘merchant silicon’ is essentially ‘not a good thing’ in the switching space. Let me clarify…Maybe an analogy will help. If a switch is like a car then the switching silicon would be most analogous to the engine and/or transmission. i.e. the core of the car and a major point of competitive advantage and differentiation. Do major automobile manufacturers outsource engine design and development to other firms? Of course not, they design and build their engines. Do manufacturers of more consumer goods like lawn mowers outsource their engines? Absolutely, they go to specialized engine manufacturers because the core value of what they offer is either a certain price point, or the value is not tied to the engine. So the question then – is do you want to ride to work or school in a car, or on a lawnmower? I know one would get me laughed at if I was in school, the other… not so much :)Applying it back to switching, I’d rather control my own destiny and align the core value creation in the silicon with the hardware and then with the software and continue to drive innovation at every tier and not saddle up on my Toro in my enterprise. (no offense to the manufacturer of lawn mowers, I am a good customer of yours too :)dg

In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.


  1. Kgraham,is there a better analogy I should have used? I am still looking for a few more good ones…dg

  2. Ori my friend, this is a great article! However, unfortunately it supports my thesis quite well. Per Wikipedia here The Fiat Group S.p.A. owns both Ferrari and Alfa Romeo. Ferrari also is the parent company of Maserati. So while they use the different brands to address different market segments and geographical sales coverage it is still one company designing and building their engine, and then, as you point out, a series of ‘tweaks’ to market it across their portfolio of products.This is analogous to us taking the switch fabric ASIC from the Catalyst 6500 which is now resident within the MDS 9500 and the Cisco Nexus 7000 as well. A perfect example of choosing the best and enjoying the best engineering out there! :)”

  3. What good is a Ferrari if it is to expensive to upkeep when it breaks down. We know all cars (and lawnmowers) will have downtimes. A lawnmower may not be as cool, but it may be higher on the dependability scale!!!

  4. My humble two cents: Aye on Doug’s comment. When comparing the value of something it helps to see whether the customer still sees”” added value directly, and whether we have reached the point of diminishing returns. As far as large Enterprise routers are concerned, we have aye on the former, and nay on the latter. Take power reduction as a simple example (since there’s a lot of green being talked here) — when done with in house silicon you can implement features, turn off I/O’s, memories, parts of ASICs, et. al. with intimate knowledge and statistics gathered online and dynamically. This is not something you can do well with off the shelf silicon. Ditto on say system/network memory architectures which need to support bandwidths of 100Gb/s. We can’t build a product which buffers 10ms at 100Gb/s with sub 2ns random cycle times without in house silicon/memory innovations/ASICs. Over time any stagnating product, for example lawmmovers where their may not be much innovation left, or extra value to the customer (some may still argue otherwise – oh well cut that noise out!), can be better addressed by off the shelf silicon (if the volumes are large enough). In contrast, we are still in the rapid innovation phase for high speed networking/data centers and in house silicon is a big product differentiator.”

  5. just a small reminder, most high-end servers used to be based on self developed RISC processors now most servers use x86 architecture, even high-end ones, since its hard to compete with Intel & AMD pace i’m not sure some of those servers can be called lawn mowers

  6. Its still a weak analogy. As we look at high-end equipment intended for reliable, long term service under heavy load, while there are always some clever exceptions, it just doesn’t hold up.Airbus, Boeing and Embraer have Pratt & Whitney, Rolls Royce and GE. For Peterbilt and International, there’s Cummins and Caterpillar.Electric Boat and Northrup Grumman, there’s Westinghouse, GE, and Bechtel. …and in some cases, Cisco has Marvell and Broadcom. There’s nothing wrong with this. The danger is in thinking that the value of the product ends with the hardware.

  7. I like this article

  8. The question is whether building your own engine results in a 1990’s minivan while taking the best engineered engine helps you build a 2008 Ferrari.See an article discussing how 3 of the top sports cars in the world Ferrari 430 Scuderia, Alfa Romeo 8C Competizione and Maserati GranTurismo all share the same V8 engine with some modifications. you choose the best, you enjoy the best engineering out there.”