perfsonar development work

[Matthew J Zekauskas <>]

I imagine route flapping sort-of looks like load balancing...

I do recall seeing load-balanced ATM links in the old Surveyor one-way
delay data.  you could see (in that case) three distinct delay bands,
that persist over time.  (This is with 30 second summaries of 2/sec
measurements, looking at min-50th %ile and 90th %-ile)  I *believe* it
was the minimum value that formed the band, but it has been a while.

If you have long-term data, then flapping would be an anomaly, visible
only in certain time frames.

Just brainstorming...

--Matt


On 7/21/2007 1:47 PM, Cottrell, Les wrote:
> Interesting question. I know of no way. Maybe someone has published 
> something on this. The only thing we looked at was looking at the 
> traceroutes and then when a change occurred looking to see whether it the 
> ASN differed, or more than the 4th octet varied, and whether it affected 
> more than one hop.
> 
> -----Original Message-----
> From: Li, Yee-Ting 
> Sent: Friday, July 20, 2007 2:32 PM
> To: Cottrell, Les
> Subject: Fwd: [pS-dev] Question about interpreting One Way Delay data.
> 
> perhaps you can answer this...
> 
> Begin forwarded message:
> 
>> From: Joe Metzger 
>> <>
>> Date: 20 July 2007 14:22:52 GMT-07:00
>> To: 
>> 
>> Cc: 
>> 
>> Subject: [pS-dev] Question about interpreting One Way Delay data.
>>
>> Hi,
>> An interesting question came up in the course of debugging a 
>> performance problem today.
>> It is not real important from an operational perspective, but it is 
>> intriguing from an academic point of view.
>>
>> Is there a way to analyze one way latency measurements between 
>> endpoints over long slow congested paths and differentiate between 
>> route-flapping and load balancing?
>>
>> If so, do you need access to raw data, or can you do it from the type 
>> of summarization that the DFN and I2 tools are typically doing today?
>>
>> Thoughts?
>>
>> --Joe
>>
> 
> 
>