Can Small Networks Keep Up?

OpenCompute1I hope this blog doesn’t come across as too negative, but it seems that every few weeks I read something that makes me think, “That is not good for small network owners.” There are a lot of changes going on at the top of our industry that are, at a minimum, worrisome for small network owners.

The biggest users of servers and related hardware have come together to create a consortium called the Open Compute Project that is working to create cheap generic hardware that will replace the expensive servers and switches bought from companies like Cisco and Juniper. This effort was started by Facebook and is likely to completely disrupt that industry. In addition to that effort, a few of the largest companies like Amazon and Google have developed their own proprietary hardware.

Recently a pile of telcos like Verizon, AT&T, Deutsche Telekom, Korea’s SK Telecom, and Equinix have joined the effort. It looks like all of the largest users of this kind of equipment will be buying their hardware from new channels, which is going to devastate the existing vendors.

One might think that this is good news for smaller companies, because in this industry small companies have always ridden the tails of the large companies. We have all benefited by having reasonable prices and a variety of option because those options were created for the big users.

But unfortunately, the Open Compute Project isn’t going to operate that way. Anybody is free to share in the open specifications that are being created, but companies are then expected to modify those specs to meet their needs and then find a way to get the gear built. Probably the top 95% of the market will no longer be buying off-the-shelf servers, which is not good for smaller users. Small companies, meaning anybody smaller then perhaps CenturyLink, will not have the resources to wade through the open source process to make their own hardware.

One might hope that there would still be somebody left to supply all of the smaller users of this equipment, but that flies against the past experience of the industry. Without big buyers of equipment, there is unlikely to be any R&D or new product development to serve a much smaller potential market. There are analysts that believe that companies like Cisco and Juniper will eventually flee the server market.

One has to worry about the general availability of telecom electronics of any kind in the future. The open source movement is not going to stop with servers; over time it will tackle fiber electronics, cable headend, settop boxes, you name it. As the big companies stop buying from vendors we are likely to see a lot of failure among the already reduced field of telecom vendors.

Along with the move to proprietary open source hardware is a similar move towards open source software to control the hardware. Again, one might think that small companies could just use the open source software, but that also doesn’t work the way you might hope. Open source software provides a sprawling mass of options and companies that use it for something like operating servers have select what they want out of it and develop their own package of options. That is way past the abilities or budgets for smaller companies. This is another area today where we benefit for the work done for larger carriers.

Small companies probably feel safe that there are a few vendors around that specialize in serving small carriers today. But many of us that are in the industry know that telecom vendors come and go. Any vendor that gets most of their revenue from the big ISPs is going to be in trouble. And when I look at the vendors used by small companies today you see almost no vendors that were here twenty years ago. The periodic downturns in the industry have always been hard for vendors to weather. There might not be enough volume from small telecom carriers to support healthy vendors for the long haul.

I hope I am wrong about all of this. Each one of these factors would be a cause for some concern. But taken all together, these trends point to a future five and ten years from now where there will be fewer vendors, where it’s going to be harder and more expensive for smaller carriers to buy gear, and where there might not be much dollar incentive for anybody to do the same R&D for small carriers that the big carriers will be doing on their own.


Leave a Reply