Looks like there are some technology changes coming that established technology suppliers won’t like. But that has been the way of the world for centuries.

One of the goals of the Open Process Automation Forum includes the idea of decoupling control hardware and software. With that in the back of my mind, this news item from one of my new favorite news sources, The Information, stood out.

Looks like the AT&T CTO is trying to do what ExxonMobil is attempting—the decoupling of hardware and software to drive down initial costs plus the costs of maintenance and upgrade. Following are excerpts from the interview of Kevin McLaughlin of The Information with AT&T CTO Andre Fuetsch. (The Information is a subscription based new media site. I don’t know if I’ve unlocked it or not.)

For big enterprise hardware companies like Cisco Systems and Hewlett Packard Enterprise, AT&T has long been a valuable customer. The telecom behemoth spends hundreds of millions of dollars each year buying devices like switches and routers that transmit data around its network. But it recently began shifting toward cheaper, less-known—or “white-box”—switches from Taiwanese manufacturers that run open-source software.

In doing so, AT&T is following the playbook of companies like Google, Amazon, Facebook and Microsoft, which run software they have written on no-name hardware. This trend has forced Cisco and other networking companies like Arista Networks and Juniper Networks to re-evaluate the way they package products. For instance, instead of selling switch hardware with software together, networking companies may have to consider selling just the software, which would hurt profit margins.


Why is AT&T making the move to white box switches instead of those made by firms like Cisco Systems?

AT&T has always had the networking expertise and capacity to do this, but we were just using [that expertise] to pick the right suppliers. We started seeing the big margins the [original equipment manufacturers] had, and how simple it was to build these boxes, and so we decided to build our own.

This has really woken up the traditional OEMs. Now they’re saying, ‘Maybe we should be in the business of not just selling a complete black box solution, but also selling our software and our hardware decoupled from each other.’

How does this decision affect longstanding relationships with suppliers like Cisco Systems?

I’m not going to comment on any specific vendor. But in general, I think it’s a really big wake-up call, and frankly, it’s going to cause vendors to change their model.

A big part of your focus these days is on “software-defined” networking (SDN), which separates high-end networking functions from hardware so they can run on cheaper hardware. At Stanford, you studied under Professor Nick McKeown, who co-founded SDN startup Nicira. What kind of impact has SDN had on the networking industry compared to what it could be in the future?

SDN has not only made networking cheaper but also more flexible—meaning you can do more things with the network, and do them more quickly.

Now the impact is getting cheaper solutions. We’ve also seen more flexibility and cycle time improvement when we develop new services. One example is mobile call recording, an application we developed for trading firms to handle Securities and Exchange Commission requirements. When you call your stockbroker and say you want to trade a stock, that voice communication has to be recorded. Before that meant the stock broker would have to take the call on their office phone. Now they can do it on mobile phones and have the recording sent back to their office recording system.

This kind of service would previously have taken us 12 to 18 months to build. But because all the network components have been turned into software, we were able to build the service in 12 weeks.

Share This

Follow this blog

Get a weekly email of all new posts.