The day software ate Cisco

Cisco's track record with software tells us the company will not survive the shift to software-defined networking

It hasn't happened yet. It's another of my predictions. But it suddenly struck me this week, as I read about the latest developments in cloud networking. Five or so years from now, the day will come when Cisco will have shrunken almost to nothing, eaten by software.

Marc Andreessen's phrase, "software is eating the world," coined for a Wall Street Journal article last August, succinctly captures the single most important trend in our world today — not only in computing but in the entire field of business. It's the driving force behind the rise of what I call frictionless enterprise — the new way of working and doing business enabled by software that leverages the connectivity of the Web plugged into the power of today's intelligent electronic devices.

At the beginning of the week, I read Cade Metz' Wired article about Nicira, under the provocative title, Mavericks Invent Future Internet Where Cisco Is Meaningless. It was an eye-opening introduction to the emergence of the new field known as software-defined networking (SDN): "a new breed of computer network that exists only as software, a network you can control independently of the physical switches and routers running beneath it ... a world where networks can be programmed like computers."

I wondered why the industry has decided to call this SDN when a much more resonant term would have been networking-as-code. It's exactly analagous to the infrastructure-as-code movement in the world of servers, powered by software like Chef and Puppet, which turns entire server farms into software-defined metadata that can be reconfigured at will. The cloud giants like Amazon, Google and Facebook have long since implemented this approach in their data centers, allowing them to run on custom specified, cost-efficient server hardware that they don't have to pay brand-name premiums to buy. Knowing that, I'm persuaded that the Wired article is spot on when it implies they're now doing exactly the same with networking, which is why Cisco faces a big challenge:

"Many of the world's largest web companies, including Google, are already buying cut-rate networking gear directly from manufacturers in Taiwan and China, making an end-run around the Ciscos and the Junipers. With Nicira providing a virtual networking platform that works with any gear from any vendors, [Nicira's CTO] Casado says, this trend will only continue. The Ciscos and Junipers, he says, will become less and less important."

Of course Cisco is aware of these developments and later in the week news emerged of its $100 million investment in SDN startup Insieme, along with an option to buy the company outright for $750 million. Om Malik gave his verdict on the leaked memo that was the source of this news under the excoriating headline, Cisco memo: We can't build anything: "when I read this memo, I see a company making a tactical admission that it has become so big, so bureaucratic and so broken that it cannot count on internal teams to build any ground breaking products."

If anything, my take is even more pessimistic than Malik's. Cisco's proven track record is that it simply can't do software. Over the years, I've become increasingly frustrated as I've watched the company flounder with a series of massive failed initatives. There was its long-forgotten venture into application-oriented networking back in 2005. Most disappointing of all was its failure to make anything much of its acquisition of WebEx.

The truth is, Cisco is a hardware engineering company and the only thing it knows how to do well is to define a need, capture it in embedded logic and then profit from dominating a locked-in market. That may still work in the client device market (as Apple continues to demonstrate) but the days are vanishing fast when you can get away with it at an infrastructure level. Software is eating that world and Cisco is on the menu. One day, we'll look round, and it will be gone.