Some pretty interesting quotes from Cisco’s CTO’s reveals that Cisco will try to leverage their dominance in enterprise networking to try and stave off the challenge of SDN. In a NetworkWorld article Cisco’s CTO is quoted as saying:
“We see the network as a platform where applications can be programmed, where information can be processed and where data and business processes can be much more efficient”
To me this sounded very similar to the overall goal of SDN. However, Cisco isn’t looking to separate the control plane from the hardware but use protocols and basically an API to allow two way control from the network and application layer. The control plane would still exist within Cisco’s hardware layer which would allow them to maintain their dominance. According to the article they are looking for enterprise partners to help drive the requirements and testing of the concept.
I think the argument for SDN still exists by leveraging the program interface as just another feature managed via a centralized control plane. Just like any other hardware based improvement is given new API’s and Management tools in an OS this programmatic network would just be an option that is abstracted by SDN. I would say that Cisco is fighting a losing battle but they have customer’s ears and could push this strategy while SDN is still forming; much like other standards they’ve pushed while the industry took time to settle.
I’d encourage you to read the article at NetworkWorld. It was very enlightening to know where Cisco is headed versus where the primary discussion has been for managing the future enterprise data center.
Maybe I’m wrong, but you could also interpret it as: not every network needs or should have centralized management like with SDN. Sometimes the application, like with the Internet of things, should configuring the network it is in ?
I don’t know, maybe they’ll just be late to the game. 🙂
Anyways I do know that if* overlay networks for virtualization really take off it could spark a race to the bottom. Because with overlay networks you need even less smarts in the network than with OpenFlow.
I’ve been very busy testing running routing protocols on the virtualization servers, if it works out it could mean even less SDN-like configuration needed and more adaptive and possibly faster failover.
* Or is it when ?
It is kind of strange if Cisco doesn’t like SDN, because SDN is like the Cloud-marketing keyword of the networking world. Which marketing department doesn’t like marketing terms ? 😉
I needed some more perspective so I was watching this one today:
https://ripe65.ripe.net/archives/video/2/
They really don’t know what to do with open network programmability. They are in denial and will defend their proprietary systems tooth and nail as they probably should. While the rest of the industry continues flocking to OpenFlow like solutions, they will resist the change and it will ultimately prove fatal.
Hmmm…. this sounds eerily similar to when Ellen Hancock at IBM denounced TCP/IP as a “research” network stating that enterprises would never abandon SNA. Later, when the rise of TCP/IP was unmistakable, IBM’s answer was to integrate it by running it over an SNA backbone.
One of the fallouts from disruptive technology is that companies and solutions we are fond of and sometimes built our career on are left behind.
The problem is, it is hard to predict which technology or technologies will ‘win’. There are no silver bullets that obviously solve the needs of people that need it.
For example, I can see certain parts of or ideas behind SDN being widely deployed. But I’m not sold on OpenFlow in hardware. It strikes me as an expensive feature. I’m not sure they’ll ever be able to make it cheap. There are many things for which that model isn’t a good fit. So it will never be the only solution, you’d still need multiple ways including the current routing protocols.
Or maybe I’m just pessimistic 😉