Re: [nmrg] Comments on autonomic use cases

[Brian E Carpenter <[e-mail]>]

Hi Joe,

On 07/03/2014 04:55, Joe Marcus Clarke wrote:
> After listening to the talks today, I have an overall comment related to
> network traceability in an autonomic environment.  I was very intrigued
> by Laurent's talk as I have been doing work in this area using Cisco
> embedded automation.  I strongly feel that we should be doing more
> autonomic cross-device analysis of problem conditions to do smart
> rerouting in the case of instability.  

Yes; and not just instability. Other conditions like impending overload
could trigger action. (Always with appropriate damping to prevent
oscillation, of course.)

> For example, in addition to local
> factors like temperature and load, we could look at path or hop
> stability in terms of dropped packets along a flow, jitter, latency,
> etc. and determine a new optimum/stable path.
> 
> Additionally, with respect to the troubleshooting section in
> draft-jiang-nmrg-an-gap-analysis-00, I think there is something that can
> be done autonomically relating to hardware failure.  Given Laurent's
> ideas again, a device that can detect imminent failure (e.g., SMART-like
> operations on harddrives), and make proactive changes to the network to
> reroute around an impending hardware failure until humans can get
> involved.  This can prevent extended outages.  Perhaps this is something
> that could supplement this section of text.

Thanks, yes. This is a case where more than one device may be involved
(which is why we're arguing that an east-west negotiation between devices
is needed).

     Brian