NTAC Peering Working Group

[Ryan Harden <>]

I doubt this was your intent, but your statement doesn't make me very warm 
and fuzzy about Internet2's IP network running over AL2S.
If none of the tools are available for the WAN and it's going to be a while 
before we have them, I'm quickly reminded of a certain cart and horse idiom. 
(Perhaps the development of these tools is already underway and they'll be 
announced alongside the migration plan for IP over AL2S?)

I share your sense that the DC is where SDN seems to be gaining the most 
traction. And IMO that seems to be where it makes the most sense, at least 
today. Traffic patterns are more predictable and many efficiencies are to be 
gained by software controlled flows for server->storage, 
storage->application, etc. 

/Ryan

Ryan Harden
Senior Network Engineer
University of Chicago - AS160
P: 773-834-5441




On May 30, 2013, at 10:05 AM, Steven Wallace 
<>
 wrote:

> IMO, for now we'll be luckly to get this feature form a single 
> controller/network. It's going to be a while before such things exist 
> across controllers. This is probably a good research topic for someone 
> (e.g., what operations API should be exposed for network troubleshooting, 
> and how to expose them). My sense is that the SDN stuff is growing up much 
> more quickly in the data center space.
> 
> ssw
> 
> On May 30, 2013, at 10:59 AM, Michael H Lambert 
> <>
>  wrote:
> 
>> Steven,
>> 
>> On 30 May 2013, at 10:52, Steven Wallace wrote:
>> 
>>> I think you've hit on something that's going to require network engineers 
>>> to re-think certain approaches to troubleshooting. IMO, the idea of an 
>>> in-band probe to determine the path is simply not going to be a part of 
>>> the SDN future (I could be wrong, but that's my prediction). The good 
>>> news is that you no longer need it. What I mean is that the controller 
>>> knows the entire path. I would be asking for an app where the engineer 
>>> can easily  say to AL2S "what's the path for this flow" and the app 
>>> responds with the complete path as it exists in realtime.
>> 
>> If there are multiple controllers, do they all contain the entire state or 
>> are they just "cooperative"?  I'm trying not to think about the old mbone 
>> traceroute that would do a hop-by-hop query.
>> 
>> Michael
>> 
>