Heavy Networking | September 18, 2015
Show 255 – Future of Networking with Douglas Comer
Ethan
Banks
Greg
Ferro
Listen, Subscribe & Follow:
What does the Future of Networking look like ? This is a special series of show where we talk with leaders in 
Networking about how they see the long term future of networking.
Todays guest is Professor Douglas Comer, Distinguished Professor at Purdue University. he was there when the Internet was a research project and part of the building it. He has published many books in networking and one
Topics
- Are autonomous routing protocols like BGP and OSPF fit for purpose for the future of networking ?
- Are overlay networks the best way forward ? Are there other choices ?
- Is Multicast still relevant ?
- Can we solve or manage the QoS problem in public or private networks ?
- Where are the skeletons buried in the TCP/IP protocol stack.
- What should network engineers be thinking about in the next decade ?
Sponsor
This week’s show is sponsored by Talari, innovator and SD-WAN leader that offers a THINKING WAN solution that proactively manages network capacity, reliability and performance so you don’t have to. Self-correcting and brilliant, Talari’s solution frees IT administrators to innovate in new area. Discover a better way to SD-WAN at talari.com. 
Share this episode
I just finished reading Internetworking With TCP/IP Volume I (6th Edition) literally three minutes before finding this post!
A super show! Gave me a good perspective. Looking forward to the next!
Great Show, FYI CDNs use Anycast not Multicast/Unicast
Some CDNs do, some don’t. At the time I was speaking internal content distribution in the CDN core.
But IP Anycast on the front end varies depending on the service offered.
Wow what a wonderful discussion! I read his TCP/IP book in college years ago. Please do more “deep dive” discussions with other professors in the future — especially CS and EE teachers. I think they will be more vendor neutral and focus more on the science, engineering and technology.
I really liked the show but it was missing a lot of stuff. The primary stuff it was missing is the future aspects of networking. Networking is meant to move packets. I like Douglas he said it all just bits. But the problem we don’t really look at the bits. But I mean really look. I can’t wait for the future of networking when bandwidth has no constraints on bits. I started my career in networking and still love it but it will disappear. Bandwidth will be like air in next 5 to 10 years. Routing protocols will be gone as well. Packet routing will also disappear. Why do I say this because we have a lot computing power now. Let me finish by saying networking is at my core but what wrong with it that we are trying to make it more stateful and we will see is it more stateless. Even under the underlay networks.
Bgp, osp and is-is will still be around when you retire
This was such a good show! I would love to see Mr. Comer as a regular guest on the podcast. It’s very interesting (and, dare I say, delightful!) to get an outlook on our field that has been tempered by the experience and wisdom he has gained in the course of his career.
Thanks so much to Greg for hosting this show!
Greg. Please let the guest talk. You are the journalist, dont clog the interview with your anecdotes and opinions. Douglas is way more interesting.
Apologies if the comment above sounded a bit harsh, not the intention. Just wanted to provide feedback to Greg. I know its tempting to jump in and share old stories and opinions on the interview, but it happened too many times and Greg was often making remarks that weren’t that relevant to what Douglas was saying, making him a bit puzzled sometimes. I would have liked less Greg and more Douglas.
Hey Luther
I appreciate the feedback. However, the process of preparing the content and recording an interview isn’t common for networking people so I often have to move things along.
Second, I had a hiccup during the edit and the two sides got out of sync. Some interruptions are the result of that.
Working on doing better.
/greg
I think that Greg style (I would say a little bit provoking) made the discussion more interesting.
I appreciate his effort to make this podcast happen. So he deserves the right to lead the discussion his way. I would love to hear another famous thinker!
Great guest and very informative interview. Greg, you get a 10 for getting Prof. Comer on the show.
Cheers
THE best PP podcast I’ve ever had the pleasure of listening to – mind blown. Ten minutes in and I was looking for somewhere to pull over so I could take some notes. I will definitely be listening to this a couple of times over when I’m not in the car. Full praise for having the humility to do this.
I’d concur with others that more academics on the show would be a good idea, the historical, scientific and (dare I say pure) engineering perspective is refreshing and highly informative. Prof. Comer literally changed my long held views of many parts of the industry and various technologies in a sentence or two.
I also think a different style of hosting is required and I’ve every confidence you’ll adapt quickly, as always.
More please.
I agree with Steven. This is the best PacketPushers podcast I’ve heard so far. Being relatively new to the field: I love to hear about the history of the technology and its transformation through the years. Especially when it is coming from guys who worked in the field and performed the research back then.
It was also really nice to hear Greg having his ass handed to him time and time again by Douglas. We all love Greg and his strong opinions and rants, but it’s also good to hear him meet his rhetorical match.
Thanks for one of the best podcasts so far! Agree with everyone else about Greg getting a good humdinger *not personal Greg*
I always wondered why all of a sudden everybody was talking about flows as if it was a ‘thing’ Doug just came along and put that in perspective for me and hopefully some others.
Regarding decoupling of endpoint from network addresses and ultimately what LISP does it would be nice to have Doug and Dino on at the same time as I thought the podcast with Dino was very convincing even though I had a hard time understanding it all sometimes.
One of the best shows I have listened to.
The best podcast ever. Especially his insight on SDN was really different from noise we hear. Can we get Mr. Doug more often?. And hope he will be allowed to talk without too many interruptions 🙂
Please invite more people like he! Great!
Best networking podcast ever. Douglas has been very patient 🙂
At the beginning of the interview ( 03:35 ) Prof. Comer speaks about some notes that I would really like to read. Unfortunately I can’t grasp the name nor the title of these notes. Could someone point me to these notes or to the author?
I belive, he meant this one: http://www.amazon.com/2014-Internet-Peering-Playbook-Connecting-ebook/dp/B00HOWTJ68/
Thank you!
William Norton also has a great site:
http://drpeering.net/white-papers/An-Internet-Peering-White-Papers-Index.html
http://drpeering.net/white-papers/Art-Of-Peering-The-Peering-Playbook.html
WOw, great source! thanks!
What is interesting (and struck me the most): I just looked into latest Comer books and have NOT found any of those insights Mr Comer has shared with us.
For example the latest “TCP/IP” books contains pure information without interesting comments on QoS (not solvable w/o having really enough bandwidth), SDN(good for researchers at this stage of development), LISP (who wants to have access to a device when the cloud is a “rendezvous point”, Role of aggregation when it comes to Internet of Things, etc.
I think we needs those insights because we, Comer books readers, believe that all presented technologies in the books are equally good;) At least when starting the career…
Another comment:
Not sure if always researchers should look into the technology so deeply as Mr Comer says when he talked about SDN. Even RIP (seems to be less complex stuff) lacked of split-horizon at first and needed this hack later – so in this case researchers failed. Maybe is a good idea to release ‘SDN’ later to test it in real life environments
I meant to release it without deep researcher thinking because even split-horizon was not anticipated by researchers…
Great guest, awesome podcast 🙂 Thanks!
If it comes to multicast use cases where one-many communication is needed please consider Public Safety Networks (Tetra for example). I mean Police or Fireman communications where multicast are heavily used for group calls. I have been working in this area for years and it seems that multicast are very effective way for this communication purpose.