Tuesday, May 24, 2011

How Does Cisco Get Out Of Trouble? OpenFlow?

MENLO PARK, CA - FEBRUARY 08:  A chainlink fen...
I'll start this entry with another disclaimer. As I said in my previous post, I'm just a guy who likes to figure things out. My current obsession of the moment is Cisco and understanding both why they got where they are today and how they could potentially find a path forward. I don't have any insider information and while my day job involves networking I wouldn't consider myself an expert. I'm currently working towards a Cisco CCNP and have a good high to mid level grasp of networking and familiarity with the major market segments. I also have a day job that is taking up a lot of my time right now and a lot of other interests. I only have a limited amount of time to research and think about this stuff.  I've never worked at Cisco and I didn't interview anyone who does before writing this and my previous entry. I really enjoy writing this BLOG though and putting words down on "paper" often helps me gain insight and a better understanding of the topics I discuss. With that out of the way...

In my previous post I covered some of the things Cisco has done that have put them in a place where their stock price has essentially been stalled for a decade plus and their competition has been eating away at their market share. I pointed out some of the downsides of a primarily acquisition based culture. Innovation suffers when you're waiting around for the next big thing. Buying external companies may save money and reduce the risks inherent in R&D but it also means you're potentially giving your competition a head start. It's easy enough to acquire small companies that come up with interesting new technologies but this is much less true if the next big thing is invented by Juniper, HP or one of the other big players.

Cisco has a lot of very smart people working for them. The tricky bit is that the company is not especially innovative. The whole culture has been built around acquisitions and their integration.

Cisco started out as an innovation company. Somehow they need to find a way to get back there. In the meantime they are going to likely be shedding unprofitable assets and continuing to refocus on their core networking expertise. They've also seen some success in the server and data center virtualization space in recent years. Servers and networking might not seem like an obvious market adjacency but given the trends brought on by virtualization they are increasingly tightly coupled in the data center. Over the past couple of years Cisco has managed to leverage that increasing integration to create a nearly one billion dollar business. That's one of the rare bright spots in an otherwise gloomy picture. Which leads me to my next point.

There has been a lot of talk about cloud computing over the past few years. Increasingly the location of data and computational resources are viewed as important only from the perspective of data security and ready availability. So long as I have access and my data is safe I don't really care where it lives. I'm not unique in this. All sorts of "magic" needs to go on behind the scenes to make this work seamlessly. Layer three TCP/IP networks are not well suited to this kind of work since they are hierarchical and inflexible as to where a particular network host/identity lives. There are things you can do with VLAN's, DNS and other technologies to mask some of this but those solutions are limited and messy. Increasingly, as Sun Microsystems used to say, "The Network is the Computer".

So one thing Cisco does have going for them is their server/virtualization products. If they can do a good job of offering solutions that overcome the issues inherent in traditional layer three networks they stand to make a lot of money. But how are they going to go about doing this? On the one hand proprietary solutions are appealing from a business perspective as they lock consumers into a complete solution. If you want all the magic you have to buy all Cisco, all the time. As a consumer, I hate that approach. I don't like getting locked into one vendor, no matter how much I like them. Companies take turns for the worse, new technologies emerge that turn the world on its head, etc. Vendor lock in is a non starter for me. Superior products and the inevitable switching costs should be enough for any well run company to keep their customers. If you have to resort primarily to lock in via proprietary features you are doing something wrong.

So how can Cisco innovate in this area quickly? One intriguing possibility is OpenFlow. What is OpenFlow? to quote the web site...

OpenFlow enables networks to evolve, by giving a remote controller the power to modify the behavior of network devices, through a well-defined "forwarding instruction set". The growing OpenFlow ecosystem now includes routers, switches, virtual switches, and access points from a range of vendors.

Notably so far as I can tell it does not include Cisco devices currently. This is a problem. The OpenFlow effort is relatively new but it smells like a game changing development to me. Cisco has been involved in OpenFlow but external appearances show them being much less interested than HP and other vendors in this market. If you're an established and dominant player in a market technologies like OpenFlow are scary because they fundamentally disrupt the way your business works. You've made assumptions and built an organization based on a certain world view. You can't just turn an organization the size of Cisco on a dime. It takes time and effort. Worst yet, how do you know when a particular trend is game changing versus being a flash in the pan? It's not an easy question to answer. Having to deal with questions like this is why CEO's and senior management get paid a lot of money.

It's my opinion that OpenFlow is going to help change fundamentally how networking, and computing is done. To be clear, I'm not saying that OpenFlow is the initiator of this change but rather that it is emerging as a very interesting enabling technology that supports trends in the industry such as virtualization, cloud computing and data mobility.

Cisco shouldn't be sitting on the side lines or conspiring to work around or subvert OpenFlow. They should be embracing what it brings to the table and leveraging their growing market presence on the server/virtualization side in conjunction with their networking business to build compelling solutions that will excite customers and investors. Cisco is well placed to carry out such a strategy and I believe it could be very lucrative. To date though they don't seem to be moving in that direction. Fighting what OpenFlow represents will be a tough battle and not one that is likely to end well. Microsoft fought the emergence of TCP/IP for several years and we all know how that ended. I think this situation is fairly analogous.

Products built solely on proprietary platforms are going to have difficulty keeping up with the innovation that a more open environment creates and fosters and even a company as large as Cisco is going to have a tough time providing all the bits and pieces that will be needed to make a purely proprietary solution work. There are just too many moving parts.

Image by Getty Images via @daylife
Enhanced by Zemanta


  1. Mike:

    An interesting post. I would like to take the opportunity to clear up some common misconceptions that are reflected in your post.

    From a market share perspective, while revenues are down, share of ports is actually pretty flat--things are not rosy, but they are not as dire as some headlines maintain.

    As far as Cisco's innovation being purely a product of acquisitions, in reality, we have one of the highest investments in R&D in the industry, both in terms of % of revenue and in terms of absolute dollars--I just did a blog post on the topic: http://blogs.cisco.com/datacenter/on-the-price-of-innovation/ Our home-grown R&D efforts have been the foundation of products like our Nexus switches and UCS unified compute platform, as well as the ability to help bring standards like FCoE, TRILL and 802,1Qbh to market. One of our core values is "no technology religion" which means we will look for the best solution where we can find it: in-house or via an acquisition. One example of this is our approach to custom ASICs vs merchant silicon: http://blogs.cisco.com/datacenter/on-the-nature-of-innovation/

    Finally, as far as OpenFlow goes, we see a lot of value in the promise of software defined networks (SDN). Rather than try and fight or subvert OpenFlow, we are members of the Open Networking Foundation. We are actively exploring OpenFlow solutions, but, sorry, no pre-announcements right now. :)


    Omar Sultan (@omarsultan)

  2. Omar, Thank you for the response. There is a lot for me to digest there.

    I think my biggest concern on the OpenFlow front is that Cisco doesn't seem as engaged as their competitors. As the established leader that is understandable but if I'm handicapping the race going forward I'm going to factor in that apparent relative lack of engagement.

    I'm certainly looking forward to finding out more about what Cisco has planned. As I said in this and my follow on post I think Cisco is in a great position to take advantage of this type of technology.

  3. On the port count front, Network hardware, particularly at the access layer is becoming a commodity and thus margins are dropping. That is not a good situation to be in. I think there are opportunities to differentiate with niche products though.

  4. Final thoughts after having read both BLOG entries and given some thought.

    The custom ASIC's take time and money to create. That isn't a criticism, just an observation. It would be interesting to know how much of that R&D spend is going to those custom ASIC's. Hardware is much faster than software and rolling your own has distinct advantages if the custom hardware implements things that customers find valuable that aren't available elsewhere but it does come with a cost in terms of time to market and investment. Hardware is likely to be less power hungry than hybrid solutions which is a plus in the data center.

    Software Defined Networks have the potential to reduce the value of that custom hardware since such technologies can be used to prototype and implement new technologies much more rapidly than is possible with customer hardware. Again, this isn't to say that there aren't advantages to the custom ASIC approach. OpenFlow is going to suffer from latency issues if the device(s) doing the management aren't topologically close to the network infrastructure they are managing.

    I wonder if another potential avenue for revenue would be building OpenFlow controllers with custom ASIC's?