RE: [RPRWG] Phy Layer question
Hi,
As Cliff suggestes, RPR should operate independent of layer 2/3 technology
and
In reference to that I would like to add few more comments:
1) Reference model
A reference model needed to show the relationship with the standard protocol
stack. Suggested reference model for 802.17:
| PMD |
| Transport |
| RPR MAC | Management Layer |
RPR MAC operations are independent of Transport layer. In addition to the
datapath interface between the Transport & MAC layers, a generic management
interface must be defined to communicate the fault & performance management
information.
2) APS
APS function can be implemented at different levels of the reference model.
There is should be enough flexibility & freedom for the system designers to
design the APS in the system based the application & system requirements.
APS can be at the PMD/Optical level or at Transport level (in the case of
Sonet/SDH, it can at line or path level) or at the MAC level.
3) Fault/Performance management
For example from the Sonet/SDH transport point of view, the Sonet/SDH path
layer performance information can be extended to RPR MAC for APS.
In the case where the transport layer does not have enough
fault/performance capabilities, RPR MAC or its related management layer must
provide necessary fault/performance management functions. At present, it is
not clear on what performance measure the RPR MAC will perform the
protection switching.
Raman
-----Original Message-----
From: Cliff Davis [mailto:cliffd@xxxxxxxxxxxxx]
Sent: Tuesday, April 24, 2001 11:52 AM
To: Wolfram Lemppenau; tak@xxxxxxxxx
Cc: stds-802-17@xxxxxxxx
Subject: Re: [RPRWG] Phy Layer question
Hi,
The Phy's should operate independent of the layer 2/3 technology, in
this case RPR. 802.17 should adopt a unique set of qualifiers that generate
alarm and switching parameters that are inclusive within the protocol and be
capable of, but not restricted to, ignoring the characteristics of the
particular phy type that happens to be utilized in the network. LOS,LOF,AIS
generated from the SONET phy could be recognized by RPR, similar to DS3 over
SONET, but also, should rely on a new set of control functions that
determine reliability of the network at the MAC layer or higher.
AIS in particular is a generated condition on detection of LOS, LOF etc.
used to deter the propagation of alarms to downstream equipment to avoid
unneeded switching and to help isolate network faults. It would not seem
possible or desirable to utilize an AIS condition in a packet based network
or protocol that depends on dynamic routing algorithms. Since RPR does not
utilize a TDM structure in the Sonet path (or over ethernet), it does not
seem possible to generate AIS to downstream MACs without declaring the
entire ring as out of service, which SONET will do anyway, if APS is used. A
completely new utility, similar to IPS in the DPT technology, would need to
be defined for 802.17.
Thanks,
Cliff Davis
Pr. Engineer
ADC The Broadband Company
8 Technology Dr.
Westboro, MA 01581
(508) 870-2506
cliff_davis@xxxxxxx
----- Original Message -----
From: Wolfram Lemppenau <wle@xxxxxxxxxxxxxxxx>
To: <tak@xxxxxxxxx>
Cc: <stds-802-17@xxxxxxxx>
Sent: Tuesday, April 24, 2001 3:02 AM
Subject: Re: [RPRWG] Phy Layer question
>
> Hi Mike,
>
> I totally agree, we need Alarm Indication Signals (AIS)
> common and independent from the physical layer. In addition
> RPR can benefit from whatever AIS is offered by the PHY.
>
> Wolfi
>
> Mike Takefman schrieb:
>
> > Vasan,
> >
> > Actually, in my opinion the debate had not yet spread
> > to LOS, LOF etc yet, but it will and should, so thank
> > you for explicitly mentioning it.
> >
> > I believe that we need to define mechanisms that work
> > for both high touch PHYs (i.e. SONET/SDH) and basic
> > PHYs (ethernet). This means the RPR MAC layer has to
> > be able to detect certain things on its own, but the
> > PHY layer can signal additional information (in theory
> > quicker and more accurately)
> >
> > cheers,
> >
> > mike
> >
> > "Karighattam, Vasan" wrote:
> > >
> > > Hi Wolfi,
> > >
> > > That is right. There are only 4 bits in the K1/K2 for src / dst
address.
> > > Only 14 of the 16 addresses are usable.
> > > But RPR has its own addressing mechanism. The debate is whether we
should /
> > > should not ignore the LOS, LOF,
> > > AIS-L, etc alarms (through SF) from sonet and replace them with new
RPR
> > > alarms.
> > >
> > > Vasan
> > >
> > > -----Original Message-----
> > > From: Wolfram Lemppenau [mailto:wle@xxxxxxxxxxxxxxxx]
> > > Sent: Monday, April 23, 2001 2:07 PM
> > > To: vasan.karighattam@xxxxxxxxx; afaber@xxxxxxxxxxxxx
> > > Cc: stds-802-17@xxxxxxxx
> > > Subject: Re: [RPRWG] Phy Layer question
> > >
> > > Hi Angela, hi Vasan,
> > >
> > > one more comment on using APS functionality of SDH:
> > >
> > > according ITU-T Rec. G.841 max. 16 nodes per ring are supported.
> > > In 802.17-rings we will have more nodes (max.).
> > >
> > > (I guess thats also one of the reasons why Cisco does not use K1/K2)
> > >
> > > Wolfi
> >
> > --
> > Michael Takefman tak@xxxxxxxxx
> > Manager HW Engineering, Cisco Systems
> > Chair IEEE 802.17 Stds WG
> > 2000 Innovation Dr, Ottawa, Canada, K2K 3E8
> > voice: 613-271-3399 fax: 613-271-4867
>