[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tsvwg] comment on draft-ietf-tsvwg-ecn-tunnel



David,

Now you've explained it, I understand your disagreement with dropping Not-ECT inner; ECT(1) outer. And you have a reasonable point - I agree. I'll change the Outgoing header in this case to Not-ECT and add reasoning along the lines of the following paragraph (I'll work on simplifying it).

============================================================================
A tunnel egress MUST clear an ECT(1) marking back to Not-ECT that must have been erroneously added to a Not-ECT packet. Forwarding rather than dropping such a packet allows this codepoint combination to be used in future standards actions. Bearing in mind this is an error case that shouldn't happen, even if forwarding as Not-ECT loses a genuine ECT(1) marking it won't be so dangerous, because any resulting lack of response to low severty marking (ECT(1)) will eventually lead to high severity marking (CE), which will be dropped at the tunnel egress.
============================================================================

Next mail I'll deal with the alarms text.


Bob

At 21:24 29/10/2009, Black_David at emc.com wrote:
Bob and Phil,

I've pulled out just the topics that I think need discussion:

> Over the last few months, our conversation has been progressing like
> two guys at each end of a blunt saw. Can we try to reach a conclusion
> on each of the three 'cuts' below? I shall be posting a rev
> before Monday.

Let's see if this sharpens the saw.  I apologize for the delays on
my part - the people who pay my salary expect me to do my day job
for some strange reason :-).  I will not be in Hiroshima (can't
justify the travel, sorry).

> 1/
> >[DB] I'm not comfortable with PCN:ECT(1) being treated like ECN:CE,
> >as I expect PCN to be setting it earlier than ECN:CE.  If the queue
> >occupancy is above the desired operating point, PCN:ECT(1) isn't
> >going to help, because it only stops the situation from getting worse
>
> [BB] You may not be aware that PCN has two threshold bit-rates: a
> lower one when PCN:ECT(1) is triggered meaning 'stop admission' and a
> higher one when PCN:CE is triggered, which means 'terminate flows'.

I have been aware of that.  PCN:CE causes reduction of traffic load in
a fashion similar to ECN:CE.  PCN:ECT(1) is a indication of less severe
congestion that is triggered earlier as congestion increase.  IMHO, it
therefore does not directly correspond to ECN:CE.

> Anyway, (paraphrasing Phil Eardley) are you uncomfortable about
> 'treatment' in the sense of tunneling treatment (wire protocol), or
> is this about PCN marking algo treatment, which ought to move
> to the PCN list?

The former.  How PCN marks packets is specified in PCN documents, and
should not be changed here.  I clearly used poor words in leaving you
and Phil with the impression that I wanted to change PCN - that was
*never* my intention (sorry).

My concern is based on the following two points:

(A) I agree with reasoning of the form "<X> needs to be done for ECN:CE,
        therefore it needs to be done for PCN:CE."  That's has to be
        valid, because if it were not, PCN would not have used the CE
        codepoint.

(B) I do not agree with reasoning of the form "<X> needs to be done
        for ECN:CE and PCN:CE, therefore it needs to be done for
PCN:ECT(1)."
        As noted above, PCN:ECT(1) is not a directly comparable
congestion
        indication to either ECN:CE or PCN:CE.

The immediate consequence of this is that we disagree on the treatment
of Inner:Not-ECT/Outer:ECT(1).  I view the arguments for Drop in this
case as instances of the (B) reasoning that I do not agree with.  Hence
I prefer the treatment of this case to be forwarding with Not-ECT (!!!)
[exclamation points indicate unusual case that gets logged/alarmed.]

OTOH, this is splitting hairs over an error case [i.e., (!!!)].  If I'm
the only person with this concern, at some point the chairs should
declare that I'm on the "rough" end of "rough consensus" and that'll
be that - this issue is not worth the time of the ADs or IESG.

> 2/
> >[DB]> >If we are going to bless that PCN usage,...
> >
> >[BB]> For Not-ECT inner the draft makes it clear:
> > >          forward ECT(0) outer; drop ECT(1) outer
> > >
> > > The question is, do you agree [to bless that PCN usage]?

Not in that fashion - see above.

> 3/
> >[DB] > >...we will need to say so clearly and bluntly in the
ecn-tunnel draft.
> >
> >[BB]> Would something like the change below be sufficient? Remember
this is
> > > in addition to all the motivation using PCN that's already in
> > > ecn-tunnel-03. Pls search thru for the string "PCN".
> > >
> > > s/ The above logic allows for ECT(0) and ECT(1) to both represent
the
> > >     same severity of congestion marking (e.g. "not congestion
marked").
> > >     But it also allows future schemes to be defined where ECT(1)
is a
> > >     more severe marking than ECT(0).  This approach is discussed
in
> > >     Appendix D and in the discussion of the ECN nonce [RFC3540] in
> > >     Section 9, which in turn refers to Appendix F.
> > >
> > > /  The above logic allows for ECT(0) and ECT(1) to both represent
the
> > >     same severity of congestion marking (e.g. "not congestion
marked").
> > >     But it also allows future schemes to be defined where ECT(1)
is a
> > >     more severe marking than ECT(0), in particular enabling the
simplest
> > >     possible encoding for PCN [I-D.ietf-pcn-3-in-1-encoding].
This
> > >     approach is discussed in Appendix D and in the discussion of
the
> > >     ECN nonce [RFC3540] in Section 9, which in turn refers to
Appendix F.
> > > /

Independent of our other discussions, I think that text change should
be made, as it clearly states the primary motivation for some of the
changes, and I see that it's been done in the -04 draft.

We have one more issue, pulling text out of an earlier message:

> FWIW, this decision about whether ECT(1) can be another form of CE
> will probably also mostly resolve the dangling topic of what
> transitions may generate alarms.

That issue surrounded what to do when the inner and outer headers
have different ECT values.  PCN uses Inner:ECT(0)/Outer:ECT(1), but
has no use for the other case, Inner:ECT(1)/Outer:ECT(0).

The log/alarm behavior in Figure 4 of the -04 draft is somewhat
alarming ;-) ;-).

Clearing ECT(1) to Not-ECT doesn't get logged or raise an alarm,
but clearing ECT(1) to ECT(0) does, while clearing CE to Not-ECT
doesn't.  Before you write a long response, I'm aware that the
explanations of what does vs. does get the log/alarm treatment
involve what happens when a new tunnel decapsulator is paired with
one of the two RFC 3168 encapsulator variants, but the result is
still strange and possibly of limited value for PCN.

Let me suggest a blunter alternative:

- Only log/alarm (!!!) by default the transitions away from Not-ECT
        in the inner header (first line in Figure 4, 3 right hand
boxes).
        These are just plain wrong, no matter what.
- Do not alarm any other combinations by default.  That puts a rapid
        end to this discussion.
- Provide an explanation (in an Appendix?) about what to alarm in a
        controlled PCN environment.  The short summary is that because
        PCN has to be deployed in a closed/controlled network, it's
        reasonable for the operator to know (in some cases) that there
        are no RFC 3168 encapsulators, and therefore be confident I
        explicitly enabling a considerably more aggressive log/alarm
        approach that catches all reduction/removal of congestion
        markings by comparison to what's in Figure 4.

If this is done, I think the current Appendix F can be bit-bucketed,
and that's probably an improvement.

Thanks,
--David
----------------------------------------------------
David L. Black, Distinguished Engineer
EMC Corporation, 176 South St., Hopkinton, MA  01748
+1 (508) 293-7953             FAX: +1 (508) 293-7786
black_david at emc.com        Mobile: +1 (978) 394-7754
----------------------------------------------------

________________________________________________________________
Bob Briscoe, BT Innovate & Design