Re: [RPRWG] Question on State Table 9.13, Receive Fairness Frame
John,
When there is no congestion on the ring, I think any sendC value less
than the number of nodes on the ring does affect the client
(#nodes <= MAZ_STATIONS).
The issue is, as I see it, that there are two different
shapers/policing signals used for generating the sendC signal.
If the hop count is incorrect, credits will be decremented from the
incorrect shaper and the bytes when added to the ringlet will be
counted in the incorrect byte counter.
If the sendC signal is set to the previous hop count value while there
is no downstream congestion, the clients tranmission past the
congestion point (that do longer exist), will be limited by the
allowedRateCongested setting and not the allowedRate setting.
Also note the last sentence in section
9.1.3.14 Computing the hopsToCongestion
...
The hopsToCongestion is assigned the value MAX_STATIONS when the local
station does not lie within a congestion domain.
-Fredrik
"John Lemon" <JLemon@xxxxxxxxxxxx> writes:
> Fredrik,
>
> I believe that the first item you mention is correct as is. Since the fairRate is FULL_RATE, the sendC indication can indicate any number of hops without effecting the client. And it seems possibly useful to retain the information about where the last congestion point was.
>
> jl
>
> -----Original Message-----
> From: Fredrik Davik [mailto:bjornfd@xxxxxxxxx]
> Sent: Thursday, October 09, 2003 2:07 AM
> To: stds-802-17@xxxxxxxx
> Subject: [RPRWG] Question on State Table 9.13, Receive Fairness Frame
>
>
>
> In row 2, when receiving a single choke fairness message from
> downstream neighboor, where:
>
> frame.fairRate == FULL_RATE
> frame.saCompact != myMacAddress
> frame.ri == myRingletID
>
> I.e. a message that that it is ok for the upstream station to increase
> its send rate.
>
> Shouldn't also the hopsToCongestion be set to MAX_STATIONS in the last
> else (in addition to setting downstreamCongested = FALSE)?
>
> Otherwise the SendC signal sent to the client is whatever the previous
> value of hopsToCongestion was...
>
> A definite error in the same row (line 13) is that rcvdSa is set to
> frame.sa. It should be set to frame.saCompact as fairness messages
> does not have a sa field (why did fairness frame have a different
> header format again ;)...
>
> Best Regards,
> Fredrik
> ---------------------------------------------
> Fredrik Davik
> Phone: +47 67 82 83 88
> Mobile: +47 45 24 91 88
> Fax: +47 67 82 82 01
> Switchboard: +47 67 82 82 00
>
> mailto:bjornfd@xxxxxxxxx
>
> http://heim.ifi.uio.no/~bjornfd
> http://www.simula.no/people_one.php?people_id=22
>
> Simula Research Laboratory
> P.O.Box 134, Lysaker
> N-1325 Lysaker
> Norway