Conference 7.286::fddi

There are assorted articles around about the ring purger, why it's good, and
why it has no impact on your network.

You're suffering from SGI FUD.  Except for about 0.1%, the ring purger does
NOT take capacity away from the ring.  It uses time on the ring that would
have been idle otherwise.

The purpose of the ring purger is to remove from the ring "no owner" frames.
There are various sources of such frames; bit errors is one, misdesigned
products is another.  Depending on their content, no owner frames can cause
SEVERE disruption of the network.  Using the ring purger will make your
network more reliable, which is why it is on by default.  

There are a few cases where it is necessary to turn the ring purger off.  If
you have nodes (bridges, typically) that violate the FDDI standard requirement
to ignore Void frames, such nodes may have trouble when the ring purger runs.
That's their fault, not ours, but even so we provide the ability to disable
the purger as a way to work around the other product's bug.

We know from empirical data that the ring purger makes a significant difference:
at one Interop show, one of the other exhibitors (a manufacturer of LAN
monitors) commented that the secondary ring had all sorts of "crud" on it,
but the primary ring was clean.  We pointed out to him that our nodes are
on the primary ring, and he was seeing the effect of the ring purger sweeping
all the junk off the ring.  "Now you know why we invented the ring purger."
He went away convinced...

Anyway, the most important message about the ring purger is that it makes
your network more reliable WITHOUT using up available bandwidth.  (For an
analogy:  Suppose you're driving down the highway, no traffic jams in sight,
traffic moving nicely.  You're getting the service you want from the highway.
It makes no difference whether there is another car behind you or not.
The void frame used by the purger is analogous to that car behind you.  It's
not slowing you down at all.)

	paul

Which adapter are you seeing this with?  The DEFZA supposedly has CNS controlling
this feature and it was supposed to be disable by default.  The DEMFA also has
this feature disabled but the driver is controlling it.    

Please let me know which adapter, the rev of code and the OS.

Thanks,

Maureen

    There might not be a ring-purger control for field-test GIGAswitch units.
    This is being corrected.  Still, you should advise the customer to
    complain to the vendor of the equipment that fails when the ANSI
    standard-conforming ring purger feature is enabled.  As Paul said, it
    is the _other_ vendor who fails to implement the ANSI FDDI standard.

    .3 Yes it is the switch and it is being fixed.
    .2 You are right, I shouldn't have said all DEC FDDI gear, I should
    have said most FDDI Gear.
    .1 Paul, Let's look at this from my customers point of view. Is the
    Ring Purge option approved by ANSI or anyone by that matter....NO!!!
    Why do we default with this on? The customer gets the feeling that" We
    know more than you Mr. Customer and although this is not approved by
    anyone we have implemented it for your benefit" This attitude is
    downright condscending. He realizes that it won't hurt his network, but
    if it is sooooo good why didn't ANSI adopt it. Are you telling me we
    know more than the standards committees??????
    
    What I am asking is did anyone ask the customer first before we did
    this? I have the largest FDDI customers in the state, (CT) and probably
    one of the largest customers in the US and guess what, the customers that
    know about Ring Purge, turn it off. 
    
    It's our damned attitude that's gotta change. All non-standard options
    should be defaulted off and if the customer wants the option he may
    turn it on. 
    
    Mike
        

Well, the ring purger IS approved by the standards committee.  Operation with
the ring purger enabled CONFORMS TO THE FDDI STANDARD.  I can quote you 
chapter and verse if you like.

I would like to hear more about "the customers that know about Ring Purge, 
turn it off."  That doesn't match my experience.  It would be useful to hear
the reasons.  If we know what the reason is we can respond to it.  (If the
reason is as simple as "I don't understand it so I don't want it" that can
be fixed simply by supplying some documentation, for example.)

	paul

    Paul,
    
    The customer knows exactly what ring purge does, over the last two
    years we have had from engineering, Bill Cronin, Cris Baldwin, Mike
    Paquette, and Carl Piper down to talk about FDDI and several times we
    have talked about the issue of Ring Purge. The customer does not see
    the value of Ring Purge. He also knows that when he looks at his
    network he sees alot of void frames on his Tecalec. (this causes
    him uneeded pain)
    
    I guess I should be more specific in my wording, the Ring Purge
    algorithm conforms to ANSI standards, but why is it an option? 
    Didn't we, (Digital) propose this to ANSI to make it part of the FDDI
    spec? Why did ANSI say thanks but no-thanks? This is what my customer
    and I have been told.
      
    It is an option, all options should be defaulted off. We got an
    attitude problem. (This reminds me of Digital's direction of SAS vs DAS 
    in the early days of FDDI.) 
    
    Mike
                  

	I think everybody who understands Ring Purger agrees that 
	technically it is a good thing and it works, but from a sales
	point of view it seems to me to be getting in the way again and
	again.

	If I was a customer, I would also want the default to be OFF.
	If I want to turn it ON I would do so on only 2 or 3 DEC units
	(probably DEFCNs on the trunk ring).  This would make it easy
	to turn OFF if any problems (real or not) come up.

	I would love to see statistics:

	1-) How many FDDI products have we sold BECAUSE of Ring Purger
	2-) How many sales have we LOST because of Ring Purger

	Luc

	

	 

Re .7: I doubt it's possible to get valid answers to those questions.

Re .6: As far as I know ring purger was not proposed to ANSI, so there wasn't
any such thing as having it rejected.  Whoever told you that doesn't sound
like someone who knows about what happened at ANSI.

The question still remains: why do the void frames cause problems?  Surely
the monitor can ignore them; packet filtering is a fundamental job of every
LAN monitor...

	paul

The customers I've talked to that complains about the void is because their 
analyzer doesn't handle them well. Ie: Sniffer and Tecalec don't seem to 
be able to disable looking at the void frames, on the other hand, my W&G DA30 
only counts the void frames but does not display them. Maybe we should get 
the customers to push back on the analyzer companies to ignore the void frames!

    
    
    Hi,
    
    I stopped and asked my customer why he doesn't like the void frames on
    the net. His answer was the added stress on the network. Granted
    everyone who works on this network from another vendor points to this
    and says "look at what DEC is doing to your network" (AT&T quote before
    they were tossed from the account) Please don't tell me that we are not
    driving the network beyond specs, or it's a good test for other vendor's
    equipment, I heard it all before. It is viewed as an unnessary stress
    on the network, not worth the benefit of what it does.
    
    Mike 

Well, ok, if you don't want me to tell you the facts but only the misperceptions
that customers want to hear, I guess I won't tell you the facts.

The fact is that's it's NOT a stress.  The cable doesn't get any warmer
from carrying void frames than it does from carrying idle symbols.  

I understand that there are cases where there are bugs and we need to work
around them by turning off the purger.  I even understand that there are
people who simply don't like the feature.  That's why we provide a way to
turn it off.  (I'm told that there is at least one product that omitted the
switch; that's clearly a defect and needs to be corrected in that product.)
But I'm NOT going to stop explaining why the ring purger is a good idea
simply because some people have an irrational dislike about hearing facts.

	paul