Threatening the market by “Wedge”? I don't think so.

A grass next to Taipei Botanical Garden (台北植物園), Taiwan

I just don't get it. Products such as iPhone/iPad are totally proprietary. And people still love them. Network devices are also proprietary. However, people just don't like proprietary network boxes and want them all to become as open as commodities?

Last Wednesday (June 18, 2014) Facebook announced the existence of their experimental in-house Top of Rack 40Gbps switch called “Wedge”. The basic ideas in many reports from the press (like this, this, and this) are: this would threaten existing market of network vendors such as Cisco.


I do not know the technical detail of Facebook’s data center. In fact, no one could ever know better than Facebook’s staffs themselves. I believe they are doing the right thing for Facebook and I really like to see if they could share the successful experience to us all in the near future.

But is the situation “threatening”? To me if the following events really happen, they are the real threats to vendors such as Cisco.


“Wedge” could be bought on the market

Yes, the products are NOT available now. To threaten the existing vendors, their products should be available on the market so they could ever threaten anyone.

Although I believe Facebook would release the success story or even the software source code, I do not think they would try to sell the combined hardware box.

By the way, could “Wedge” be cheaper than existing switches in the market? Let’s move on to my next point.


“Wedge” could be built with commodity hardware components

Published Wedge Hardware Design.
Source: captured on Facebook Engineering Blog.

A key components I saw in the following diagram of “Wedge” is a 16-port-40Gbps Ethernet server adapter. I do not believe this hardware would be easily available on the market. Even so, it should not be cheap because the sales volume should be far from that of other type of adapters.

This is also true even if the key adapter were a 16-port-10Gbps Ethernet adapter.

It is a luxury, not a commodity at all.


“Wedge” could be easily assembled, tested, installed, and easy to troubleshoot.

Assembling, testing, installing, and troubleshooting are all demanding huge man hours to do. Unless “Wedge” is so simple in all activities that customers do not have to hire many additional staffs to do the extra workload, it might not be a cost effective option for most of the customers I know of.


[My Conclusion]

I DO like open source. I also dream of a world where everything is just open. But I also hate to fail on the business targets because I choose risky solutions.

An available and working solution is much much better than anything else! Don't you think so?


[References]

Introducing “Wedge” and “FBOSS,” the next steps toward a disaggregated network | Engineering Blog | Facebook Code