[Rstplib-users] RE: 802.1w and timer dependencies

[<al...@nb...>]

On Sent: Wednesday, November 21, 2001 9:54 AM Mick Seaman wrote:

Mick > Re: your question on why the  .1w handshake is better than 
Mick > the handshake in
Mick > backbone fast.

I didn't ask it at all (oh, my poor English !) !
I propose to extend this (finest and rapidest !) handshake of .1w in
'upstream' direction. I would like to use the same spirit, logic, speed and
timers independence as .1w handshake !

Nevertheless, it look after Mick's postings, that we needn't do it, because
there is no the problem with "indirect link fail" in 802.1w.
Let us return to this discussion after this question resolving.

Best wishes, Alex


> -----Original Message-----
> From: Mick Seaman [mailto:mic...@ie...]
> Sent: Wednesday, November 21, 2001 9:54 AM
> To: Ar...@op...; std...@ie...; 
> rst...@li...; bri...@ie...
> Subject: 802.1w and timer dependencies
> 
> 
> 
> Alex,
> 
> Re: your question on why the  .1w handshake is better than 
> the handshake in
> backbone fast.
> 
> In .1w timers are only used to recover from lost mesages on 
> point to point
> links to bridges/bridge ports that are functioning, in other 
> words the speed
> of recovery does not depend on the timers (except in cases 
> where the maximum
> message transmit limit kicks in). In back bone fast there are 
> messages to
> test if a bridge is "still there", which means that some (not all)
> recoveries have to wait a time for non-response before taking 
> appropriate
> action. This introduces a timer which in its normal case of 
> operation can be
> "tweaked" to improve user performance - which opens the door 
> to overzealous
> tweakers.
> 
> All timers are evil, but some are more evil than others.
> 
> Mick
> 
> -----Original Message-----
> From: Alex Ruzin [mailto:al...@nb...]
> Sent: Tuesday, November 20, 2001 10:54 PM
> To: mic...@ie...; std...@ie...;
> rst...@li...; bri...@ie...
> Subject: RE: Question: 'upstream' handshake in 802.1w
> 
> 
> Mick,
> 
> I don't understand you: what do you mean under "this certainly
> works". Yes, in .1w the topology becomes stable, fully and 
> simply connected,
> but after only about 6 seconds.
> 
> [Apropos, it seems to me, that the same problem is an explanation
> for case of http://www.ieee802.org/1/private/email/msg00348.html]
> 
> My proposition is devoted to decrease this time.
> 
> Why such handshake is critically timer dependent ? Why it is worse
> than the handshake, described in the second half of 17.19 and
> using variables like 'proposing', 'propose', 'sync', 'synced' 
> and 'agreed' ?
> 
> I propose *in addition* to introduce new and similar variables and to
> use new bits in the field "Flags" of RSTP BPDU.
> 
> Your reply will be accepted with gratitude.
> 
> With respect, Alex
> 
> > -----Original Message-----
> > From: Mick Seaman [mailto:mic...@ie...]
> > Sent: Wednesday, November 21, 2001 4:06 AM
> > To: 'Mick Seaman'; Ar...@op...;
> > std...@ie...; rst...@li...;
> > bri...@ie...
> > Subject: RE: Question: indirect link fail in 802.1w
> >
> >
> >
> > This certainly works (subject to accuracies of representation
> > in the spec of
> > what "this" is) without such a handshake (see you closing
> > statement below),
> > and as the handshake is critically timer dependent it is
> > against the spirit
> > of .1w. I was aware of the handshake mechanism well before
> > deisgning .1w.
> >
> > Mick
> >
> > -----Original Message-----
> > From: Mick Seaman [mailto:mic...@ie...]
> > Sent: Tuesday, November 20, 2001 5:14 PM
> > To: 'Ar...@op...'; 'std...@ie...';
> > 'rst...@li...'; 'bri...@ie...'
> > Subject: RE: Question: indirect link fail in 802.1w
> >
> >
> > Alex,
> >
> > I suggest you test this out using the RSTP Visio simulation.
> >
> > The only case in which any of these ports could be Backup is
> > a deficiency in
> > the spec or the interpretation of the spec, so extra protocol is not
> > required.
> >
> > Mick
> >
>