[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Optical impairments in draft-ietf-mpls-generalized-signaling- 04
-----Original Message-----
From: Heiles Juergen [mailto:Juergen.Heiles@icn.siemens.de]
Sent: Wednesday, June 27, 2001 12:19 AM
To: 'John Drake'; Heiles Juergen; 'Yangguang Xu'; Diego Caviglia
Cc: Ayan Banerjee; ccamp@ops.ietf.org; Giovanni Fiaschi
Subject: RE: Optical impairments in
draft-ietf-mpls-generalized-signaling- 04
John,
wavelength availability and wavelength conversion are two different points
in my view.
A optical corss-connect without wavelength conversion can be seen as a
connection matrix with limited connectivity (e.g. only connections between
inputand output ports with the same lambda are possible), while a
corss-connect with wavelength conversion has full connectivity if no other
constrains apply. This is a static information that doesn't change over
time. It is detected during startup of the network and will not result in
flooding traffic.
JD: Good point, and as I indicated in my note, today we're implicitly
assuming that every link is capable of wavelength conversion and has no
impairments. Perhaps a note to this effect should be added to the routing
draft. (In general, I think the plan was to assume as a default that links
were unimpaired and that links with impairments would be advertised as
exceptions.) However, wavelength conversion capabilities are not advertised
between areas or ASs, so signalling still has to deal with wavelength
continuity. Further, in the future wavelength conversion may be a node
based limited resource, both wrt capacity as well as capability (i.e., which
input frequency can be converted to which output frequency), so it probably
won't be static and advertising it could cause a flooding load. And, again,
it doesn't help between areas or ASs and there are race conditions so
signalling still has to deal with wavelength continuity.
The available connectivity of a connection matrix can be taken into account
by the routing.
Wavelength availability is a second issue. If you want to annouce the used
and unused wavelengths in real-time you end up with flooding traffic and
race conditions, I agree. But this is independent of wavelength conversion
availability. A label set helps in my view only if you have wavelength
conversion as it allows you to select between different wavelengths.
JD: Peter and Lou are the experts, but I think that the point of label sets
was to allow you to discover points along the path with wavelength
conversion capabilities as part of signalling.
Without wavelength conversion you are stuck to a fixed wavelength. What you
need is a cranck back mechanism if you detect that all possible wavelengths
are occupied.
The availabiity of wavelength conversion and the connectivity of a
connection matrix as such is a separate issue from PMD or OSNR.
Note that even in TDM systesm (SONET) you may have limited connectivity.
Some SONET ADMs for example don'tsupport time slot interchange for the
line-line connection. That results in the same connectivity restricitions as
an Optical ADM without wavelength conversion between the line interfaces.
We should separate the issue of wavelength conversion (-> limited
connectivity) and signal impairments.
Wavelength conversion usually has impact on signal parameters, but that can
be modelled as a separate process wich can also exists without wavelength
conversion.
Today wavelength conversion ist mostly based on o-e-o conversin with
re-timing (3-R). The 3-R process resets the otpical signal parameters, but
it influences timing parameters (jitte, wander).
Wavelength conversion based on o-e-o conversion with re-shaping (limiting
amplifier), but without re-timing is also available (2-R). This has
influence on other parameters.
Pure optical wavelength conversion will have other influences. We don't know
exactly which today and it may depend on the specific process used.
Consideration of signal impairments in the routing process is a much more
complex issue compared to limited connectivity if I look on the tools we use
today to design an optical path.
JD: I think that we might be agreeing with each other; i.e., impairments
should be handled in routing and wavelength continuity should be handled in
signalling.
Juergen
> -----Original Message-----
> From: John Drake [SMTP:jdrake@calient.net]
> Sent: Tuesday, June 26, 2001 11:58 PM
> To: 'Heiles Juergen'; 'Yangguang Xu'; Diego Caviglia
> Cc: Ayan Banerjee; ccamp@ops.ietf.org; Giovanni Fiaschi
> Subject: RE: Optical impairments in
draft-ietf-mpls-generalized-signaling- 04
>
> Announcing wavelength availability information in routing doesn't work
> because of the flooding traffic it would generate, plus there are still
race
> conditions that would have to be handled by signalling anyway. Label set
> was intended to support wavelength continuity issues.
>
> Also, in the short term, we're dealing with transponder based interfaces
to
> the DWDM systems so this is a non-issue. In the longer term, this is just
> another issue like PMD and OSNR that will need to be dealt with.
>
> -----Original Message-----
> From: Heiles Juergen [mailto:Juergen.Heiles@icn.siemens.de]
> Sent: Tuesday, June 26, 2001 7:43 AM
> To: 'Yangguang Xu'; Diego Caviglia
> Cc: Ayan Banerjee; ccamp@ops.ietf.org; Giovanni Fiaschi
> Subject: RE: Optical impairments in
> draft-ietf-mpls-generalized-signaling- 04
>
>
> Yangguang,
>
> wavelength conversion is in my view an issue for path selection as you
> cannot select a certain path if the wavelength doesn't fit and you have
no>
> wavelength conversion.>
>
> Regards
>
> Juergen
>
> > -----Original Message-----
> > From: Yangguang Xu [SMTP:xuyg@lucent.com]
> > Sent: Tuesday, June 26, 2001 3:13 PM
> > To: Diego Caviglia
> > Cc: Ayan Banerjee; ccamp@ops.ietf.org; Giovanni Fiaschi
> > Subject: Re: Optical impairments in
> draft-ietf-mpls-generalized-signaling-04
> >
> >
> > Diego,
> >
> > As indicated by the document that Ayan pointed to you, there are many
> constrains
> > for the optical networking. These constrains can be considered at
network
> > planning time, path selection time (routing using OSPF-TE/ISIS-TE) or
> connection
> > setup time(GMPLS signaling). When a constrain is considered is an
> engineering
> > issue. We just have to figure out when and where is the best and
cheapest
> to
> > consider a constrain. To wavelength conversion, GMPLS signaling is a
good
> place
> > to handle.
> >
> > Thanks,
> >
> > Yangguang
> >
> > Diego Caviglia wrote:
> > >
> > > Ayan,
> > >
> > > Thanks for your answer but I don't understand why lambda
> conversion is
> > > a signalling problem and thus is covered in generalized signalling
> draft while
> > > others optical impairments are routing problem and are covered in
a
> separate
> > > document.
> > >
> > > Moreover even if the document you quoted is a well written and very
> useful one
> > > it is an informative contribution that simply points out the
> physically
> > > impairments of an all optical network. It doesn't propose any
> extension to
> > > OSPF/IS-IS in order to support optical routing.
> > >
> > > Given that in an All Optical Network we have to cope with
> lots of new
> > > information, as you stated in your document, are we sure that
OSPF,
> in this
> > > environment, scales?
> > >
> > > I mean if it doesn't scale for wavelength availability information
> distribution
> > > why it has to scales for other optical impairment distribution?
> > >
> > > ----------------------------------------------------------------
> > > Diego Caviglia
> > > Photonic Networks Design and Modelling
> > > E-mail: diego.caviglia@marconi.com
> > > Tel: +39 (0) 10 6003 808
> > > Via A. Negrone 1A 16153 Genoa (Italy)
> > > http://www.marconi.com
> > >
> > > Ayan Banerjee <abanerjee@calient.net> on 25/06/2001 22.26.02
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > To: Diego Caviglia/MAIN/MC1@MCMAIN,
> > > petera@nortelnetworks.con, ccamp@ops.ietf.org
> > >
> > > cc: Giovanni Fiaschi/MAIN/MC1@MCMAIN
> > >
> > >
> > >
> > > Subject: RE: Optical impairments in
> > > draft-ietf-mpls-generalized-signaling- 04
> > >
> > >
> > >
>
----------------------------------------------------------------------------
> ----
> > >
> > > Diego,
> > >
> > > Please see draft-ietf-ipo-impairments-00.txt for optical constraints.
> > >
> > > Thanks,
> > > Ayan
> > >
> > > -----Original Message-----
> > > From: Diego Caviglia [mailto:Diego.Caviglia@marconi.com]
> > > Sent: Monday, June 25, 2001 6:19 AM
> > > To: petera@nortelnetworks.con; ccamp@ops.ietf.org
> > > Cc: Giovanni Fiaschi
> > > Subject: Optical impairments in
draft-ietf-mpls-generalized-signaling-04
> > >
> > > Peter ,all,
> > >
> > > a comment about
draft-ietf-mpls-generalized-signaling-04.
> In
> > > the
> > > draft there is the label set object that is very useful in All
optical
> > > networks
> > > with CI-incapable nodes but lambda conversion is only one of
the
> > > problems
> > > related to path establishment in such networks. I mean what
about
> OSNR
> > > and
> > > non-linear impairments? What about gain variation and power
> equalization?
> > > Where
> > > are covered these topic?
> > >
> > > If I can find a path feasible from the lambda continuity point
of
> view
> > > who
> > > assure me that I have the needed Q at the end of the all optical path?
> > >
> > > Best regards.
> > >
> > > ---------------------------------------------------------------->
> > > Diego Caviglia>
> > > Photonic Networks Design and Modelling
> > > E-mail: diego.caviglia@marconi.com
> > > Tel: +39 (0) 10 6003 808
> > > Via A. Negrone 1A 16153 Genoa (Italy)
> > > http://www.marconi.com