RE: [Sipping] Sip Performance Metrics: Average Hops per INVITE (AHI)and Average Hops per Session (AHS)

[Daryl Malas <[e-mail]>]

Adam,

I know there have been various items within the working group to detect
route loops, such as the draft you described below.  I think of the
number of 483's as a fault, based on a predefined threshold.  IRT, ISO
FCAPS, the difference lies in a Fault condition versus a Performance
metric.  My draft is attempting to define Performance metrics, instead
of detecting faults...that being said, you could predict or detect a
potential fault based on evaluating the output from a performance
metric. ;-)

Thanks...--Daryl

On Tue, 2006-06-20 at 22:03 -0400, Uzelac, Adam wrote:
> Daryl,
> 
> This email has been in my queue for numerous days.  Sorry for the delay.
> I think that establishing SIP performance metrics is a very valuable
> exercise, and I appreciate the effort. Now as for the topic at hand, and
> specifically the "Hops per x"....Outside of the AHI Tuna reference that
> splashed into my head ;) here are some thoughts.  
> 
> Your looking into quantifying hop detection is congruent with recent
> focus on draft-ietf-sip-hop-limit-diagnostics-03. Detecting the
> existence of a loop is basically knowing a 483 has been returned. Right?
> Why not just determine the frequency of 483 Too Many Hops messages?
> draft-ietf-sip-hop-limit-diagnostics would help in determining the
> culprit.
> 
> With that said, the metrics you put forth blow might be valuable in
> determining propensity to the equivalent of post dial delay.  Just a
> thought.
> 
> Adam
> 
> -----Original Message-----
> From: Daryl Malas [mailto:[e-mail]] 
> Sent: Friday, June 16, 2006 12:37 PM
> To: [e-mail]
> Subject: [Sipping] Sip Performance Metrics: Average Hops per INVITE
> (AHI)and Average Hops per Session (AHS)
> 
> I have had a request to add a metric, which could help avoid routing
> loops in an operators network.  I have come up with two metrics, and I
> wanted to get feedback from the mailing list before I update the draft.
> Here is what I'm thinking:
> 
> 
> > Average Hops per INVITE (AHI)
> > 
> >  
> > 
> > AHI is calculated as an average and is defined as the number of hops 
> > per INVITE request.  This metric is used to indicate potential 
> > inefficient routing and/or help an operator detect and/or prevent 
> > routing loops.
> > 
> >  
> > 
> > Variables =
> > 
> >             a = # of INVITE requests
> > 
> >             b = SUM of a "Max Forwards" value
> > 
> >  
> > 
> >                       (a * 70) - (b)
> > 
> >             AHI = ----------------------
> > 
> >                              a
> > 
> >  
> > 
> > The value of 70 is used based on the recommended origination value for
> 
> > each request as described in section 8.1.1.6 of [2]. Any consistent 
> > value used by an operator or provider may be used in place of the 
> > recommended value.
> > 
> >  
> > 
> > Average Hops per Session (AHS)
> > 
> >  
> > 
> > AHS is calculated in a similar manner to AHI; however, the "Max 
> > Forwards" value is taken from each request associated with the entire 
> > session as described in the following section 2.5.  This metric is 
> > also used in a similar manner as AHI.
> > 
> >  
> > 
> > Variables =
> > 
> >             a = # of SIP requests
> > 
> >             b = # of completed sessions
> > 
> >             c = SUM of a "Max Forwards" value
> > 
> >  
> > 
> >                     (a * 70) - (c)
> > 
> >             AHS =---------------------
> > 
> >                           b
> > 
> >  
> > 
> > The value of 70 is used based on the recommended origination value for
> 
> > each request as described in section 8.1.1.6 of [2].  Any consistent 
> > value used by an operator or provider may be used in place of the 
> > recommended value.
> > 
> Your feedback and thoughts are appreciated.
> 
> Thanks...
> 
> --Daryl
> 
> 
> _______________________________________________
> Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
> This list is for NEW development of the application of SIP Use
> [e-mail] for questions on current sip Use
> [e-mail] for new developments of core SIP
> 
> _______________________________________________
> Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
> This list is for NEW development of the application of SIP
> Use [e-mail] for questions on current sip
> Use [e-mail] for new developments of core SIP


_______________________________________________
Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use [e-mail] for questions on current sip
Use [e-mail] for new developments of core SIP