Lather, Rinse and Repeat
I was reading some recent SDN announcements from various networking companies. When I read about mediating between OpenFlow requests and traditional control planes like a session border control device as a SDN solution, my reaction is to just shut down. I no longer accept input. If you are talking about retrofitting traditional network designs to accept OF commands then you are missing the point of the SDN movement. That is why I wrote the post on missing the point of the SDN movement. There are two camps in the SDN movement and most people are in the “let’s keep doing what we have been doing, but provide some input mechanism for OpenFlow” camp. This is the group of people who ultimately do not want the network to change.
Anytime you here a presenter say “the network is fine, the network works don’t touch it” and “new networks cannot scale, we tried all this before and it did not work, what we have today works just fine” then you know you are talking to a person who does not want the network to change. Change makes them uncomfortable. To combat change they wrap their existing network solution in a bubble of SDN terms and call it “a powerful, revolutionary movement that will transform your network at the forefront of change.” Unfortunately, these are the people who want to continue to build networks the same way (i.e. 2-3 tier networks (CLOS), distributed protocols, traffic engineering, QoS, DPI, etc) and they are only addressing the SDN movement through the OpenFlow blow hole method. They want some level of OpenFlow interaction with their control plane software, but their ultimate goal is extend the life of switched network design for another 15-20 years, hence building off their base. Build it wider, build it higher and pile some more software on top. To get customers to upgrade the network, a serious amount of SDN washing is needed.
I call this the Lather, Rinse and Repeat group. They want to build the same network they have been building since 1992 and the SDN or OpenFlow movement is just a nuisance. This is the creativity fail group. The SDN movement is a threat to the switched network architecture. To get customers to upgrade their switched network to the next level of switched networks their solution is to lather up existing solutions with an ample amount of SDN, rinse and repeat. My rant is at an end and as always I could be mistaken.
Reminds me of South Pacific where she sings: “I’m gonna wash that man right outta my hair”
True, SDN requires the network to change. But there’s a balance of not forcing people to rip & replace all their management stuff, otherwise nobody will ever adopt SDN. So we should build the radically new networks first, and adapt them to the legacy tools, not the other way around.
Shameless plug alert: for more about how SDN is a transformative technology, see my blog on the Open Datacenter Interoperable Network (ODIN) https://www-304.ibm.com/connections/blogs/DCN/entry/odin_sets_the_standard_for_open_networking21?lang=en_us
or my tweets on the same topic: https://twitter.com/#!/IBMCasimer
I see it as cap and grow. The new networks start off to the side and interoperate with the legacy networks. 2012 is the year that proof of concepts start and evolve into 2013. Next year is the year of the big changes when the world realizes that multi-tier switched networks of the last era are passé. Nice to have a comment from you Cas…I read the ODIN (all four parts) over the weekend. Shameless plug by me too…Plexxi news at GigaOM Structure.