Measuring PoP to PoP latency--tools to use?

Mailman List Mirror (Read Only)
1

### On Wed, 22 Aug 2001 17:38:25 -0700, k claffy <kc@ipn.caida.org> casually
### decided to expound upon nanog@nanog.org the following thoughts about
### "Re: Measuring PoP to PoP latency--tools to use?":

how many ISPs use this and [how] do they
find it helps them operationally?

I don't know any ISPs that us it. I imagine some enterprise networks do. I
played around with it in anticipation for using it to measure jitter but
found it of limited usefulness because if you have to deploy a box to do
localised collection of the data anyways you may as well run a better jitter
probe from that unix box.

it has been around for years, during which time
i have heard no actual data wrt its operational utility,
reckon that cisco would be responsive
to ISP suggestions for improvements to it

I thought I read somewhere that Cisco was planning on dropping RTR from the
featuresets. Then again, I may have just imagined that.

(if other router vendors have similar functionality
they should probably speak up;
we can put such features in the caida tool taxonomy)

Seconded...

maybe a nanog panel on provider experiences/caveats
on any of such vendor-provided functionality
is not a terrible idea.
mod all the anxiety involved in sharing such experiences

I remember at one time that on-the-router measurement features were kinda
frowned upon. At least that's the general feeling I got from people. Has
this attitude reversed?

2

I will try not to violate my NDA here but there is a company that is nearly
finished beta testing a service (hardware based) that is designed to measure
latency, delay, jitter and packet loss. The initial product is meant
specifically for the core but there is an edge product in development as
well. I can't tell you anything more without getting myself into trouble.

Regards,

Larry Diffey

3

I'll take the opportunity to plug some old friends at the RIPE Test
Traffic project. They are measuring and analysing unidirectional latencies
and routing vectors using GPS as a precision timesource. Daniel Karrenberg
shed some statistical light on their raw data at RIPE-38.

http://www.ripe.net/test-traffic/ for more information.

Joshua