[Nfd-dev] [Operators] monitoring data plane

[Junxiao Shi]

Dear folks

An alternate is: write a program to connect to each HUB via ndn-cxx's
TcpTransport, and send ping Interest over there.
Benefit: This program needs to run on only one machine. No deployment on
HUBs.
Drawback: RTT includes the delay between the laptop running this program
and the HUB - but it's sufficient to test reachability.

Please check out the reachability page for my ndn6.tk node:
http://yoursunny.com/p/ndn6/reachability/
It uses WebSocket connections to every router.
It works on recent version of Chrome, including desktop and Android
editions.
"NDN6" and "UCLA/v6" will appear red if your laptop doesn't have IPv6.

Yours, Junxiao

On Sat, Aug 16, 2014 at 8:03 AM, Beichuan Zhang <bzhang at cs.arizona.edu>
wrote:

> Hi John,
>
> The current testbed status page monitors the control plane, i.e., prefixes
> in the routing table. I think we also need monitor data plane reachability
> e.g., measured by ndnping.
>
> More specifically,
>
> - Each node runs a script that reads the list of prefixes from a local
> file, and ndnping each prefix for a number of times, say 10. The output is
> published as an HTML or XML page under the current web server on the node.
> The measurement is done periodically.
>
> - The stat collection script at WashU retrieves the page from individual
> node, parse them, and the result is another table to the current status
> page. This table has the same layout as the current table, but in each cell
> it shows the % of ping packets received in the most recent measurement.
>
> This information would be helpful once apps are deployed on the testbed.
> What do you think?
>
> Beichuan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/nfd-dev/attachments/20140816/efca4ca5/attachment.html>