Re: [T1X1.5] Re: Suppression of Downstream Alarms...

[Carmine Daloia <[e-mail]>]

Jonathan,

Actually the OTN involves more recommendations than G.709 (e.g., G.872, 
G.798, etc.). That is why I didn't specifically call out G.709. However, 
it seems that more people are familiar with the term G.709 rather then 
standard OTN and therefore I will make the changes you suggest. I think 
this should be okay, especially since G.709 itself refers to other 
recommendations and therefore by pointing directly to G.709, implicitely 
points to other recommendations as well.

Thanks
Carmine

Jonathan Lang wrote:

>Carmine,
>  To be precise, when you refer to "standard OTN" and "non-standard DWDM",
>are you referring to G.709?  If so, I would suggest you do a global
>search-replace of your proposed text to read "G.709" and "non-G.709",
>respectively.
>
>Thanks,
>Jonathan
>
>-----Original Message-----
>From: Carmine Daloia [mailto:[e-mail]]
>Sent: Wednesday, November 21, 2001 9:51 AM
>To: Sudheer Dharanikota
>Cc: Jonathan Lang; [e-mail]; [e-mail]; [e-mail]
>Subject: Re: [T1X1.5] Re: Suppression of Downstream Alarms...
>
>
>Hi Sudheer,
>
>My last comment wasn't meant to be an argument :-)
>
>Maybe if I propose some text specific to the suppression of downstream
>alarms that would help clarify the scope of applicability. The text would
>state that when client devices (e.g., SONET/SDH cross-connects, IP routers)
>are interconnected via a standard OTN network then the suppression of
>downstream alarms is already handled in the transport/user plane via the OTN
>overhead (both in-band via the digitial overhead as well as out-of-band via
>non-associated overhead). Also the text would address PXCs within a standard
>OTN network. In this case, again the suppression of downstream alarms is
>handled via the OTN overhead.
>
>The implementation proposed in LMP for suppression of downstream alarms
>applies to PXCs or client devices (e.g, SONET/SDH cross-connects or IP
>routers) interconnected via a non-standard DWDM network. In this case, it is
>assummed that the non-standard DWDM network does not provide the neccesary
>overhead within the transport/user plane to suppress alarms on PXCs and
>client devices and therefore LMP provides a mechanism to carry such alarm
>suppression messages in the control plane.
>
>I'll take a crack at specific text so that the group can review it. Does
>this sound like something that would be helpful.
>
>Thanks
>Carmine
>
>Sudheer Dharanikota wrote:
>
>Hi again: 
>Carmine Daloia wrote: 
> 
>I am not assuming inband communication is supported between the
>cross-connect and the line system. That is why I said that when the PXC and
>line system are from different vendors such OAM signals would have to be
>carried over a separate channel (e.g., LMP-WDM control channel). However, I
>don't see why such signals would have to be carried over an LMP control
>channel between cross-connects. Also, my point about performance was that we
>would have to see if the LMP control channel would meet the performance
>requirements for such OAM signals. It very well might. If not, a
>bit-oriented signaling interface may be needed between the PXC and the line
>system. 
> 
>Ok Let us address the issues separately... 
>1. Communication between Line System and Cross connect 
>Defining a bit-oriented (physical) interface and protocol is not really the
>task (in my opinion) 
>of IETF. 
>I am not sure if are creating a problem to solve with the alarm suppression.
>
>Let me explain...
>Let us assume we can correlate and suppress the alarms between the Line
>systems 
>and corss connects. Then we donot need an elaborate bit-oriented protocol.
>This 
>has an assumption that we are talking about not hundereds of failures but a
>couple of 
>failures (for example fiber cuts).
>As I said, I am not suggesting that we need a bit-oriented protocol. My main
>point was that for the communication between the line-system and
>cross-connect, we need to understand the applications that require such
>communication and the requirements for those applications. It may be that
>the requirements can be met by carrying messages over IP (e.g., LMP-WDM).
>This is an issue for LMP-WDM and not for LMP (i.e., running between
>cross-connects) anyway. So for the sake of focusing some of this discussion
>let me suggest we focus on communication between the adjacent cross-connects
>and we can even assume that LMP-WDM handles the Cross-connect and Line
>System communication.
>Agreed we need to focus on LMP not DWDM-LMP :-) But... 
>the intention of the OLI document was to underdstand the requirements. 
>I thought most of us vouch for these requirements. Let us put the 
>requirements in the back-burner. If it is bit-oriented protocol it 
>does not interest most of us :-) 
>  
>2. Communication between cross connects 
>THis is needed to localize a fault which in turn can be used as a trigger
>for local 
>restoration.
>Localize a fault for what reason? To understand if the fault is within the
>protection/restoration domain. For span protection/restoration the
>end-points of the domain coincide with the LMP end-points. For end-to-end
>protection/restoration the end-points do not coincide with the LMP
>end-points. Because the end-points of a protection/restoration domain do not
>always coincide with the end-points of a LMP protocol, I don't think it is a
>good idea to lump messages related to protection/restoration in LMP. It will
>result in similar messages having to be defined in other protocols and
>therefore a duplication of work and processing. 
> 
>Again... span protection can be performed easily with the current 
>mechanisms supported by LMP in out-of-band communication 
>scenarios. 
>In case of overlaping restoration scenarios such as span and end-to-end 
>it is required to have a localization + an attempt to recover + (if this
>fails) 
>report to end-to-end entities for recovery. THis also vouch for the 
>inclusion of such mechanisms in LMP. 
> 
>Will this duplication cause major issues in a network. I doubt it... but it
>just doesn't seem right from an architecture perspective. 
>
>Well I guess I cannot buy this argument :-) 
>Cheers, 
>sudheer 
> 
>  
>- sudheer 
> 
>Thanks 
>Carmine 
>Sudheer Dharanikota wrote: 
>Carmine... 
>Again can you suggest a way to communicate such signals 
>between line systems and cross connects when inband 
>communication is not supported between these two systems? 
>Cheers, 
>sudheer 
>Carmine Daloia wrote: 
>Jonathan, 
>Forgot to mention, that the performance aspects of carrying OAM type signals
>over an IP based control channel in LMP-WDM would have to be analyzed. It is
>possible that the IP Control Channel will not provide fast enough transfer
>to actually suppress downstream alarms, however that needs to be analyzed as
>part of LMP-WDM. 
>Thanks 
>Carmine 
>Carmine Daloia wrote: 
>Jonathan, 
>The LMP-WDM document specifies the signaling between the Cross-connect and
>OLS, assuming they are from different vendors. If they are from different
>vendors, then a standard interface is needed to exchange some information.
>One type of information that would need to be exchanged is some OAM signals.
>Maarten described some of these signals in his VBI document. However, I
>don't see why OAM signals would have to be exchanged directly between the
>cross-connects themselves via LMP. 
>Let's look at the following network. 
>OXC1 --- OLSA --- OXC2 --- OLSB --- OXC3 --- OLSC --- OXC4 
>Note that the OLS consists of DWDM Mux/Dmux Terminals and Optical
>Amplifiers. 
>Let's assume a failure on OLSA. OLSA via overhead within an OSC suppresses
>alarms within OLSA. OAM messages (e.g., Optical Channel FDI) could be
>carried over the LMP-WDM control channel to OXC2. OXC2 will have to forward
>the FDI signals downstream over the LMP-WDM control channel to OLSB. OLSB
>will then forward these FDI signals over its OSC and then over the LMP-WDM
>control channel to OXC3..... etc... 
>Note that OXC2 does not need to directly forward these FDI signals to OXC3.
>So it is possible, that in LMP-WDM, we may need to define messages
>corresponding to FDI signals to suppress downstream alarms, however we don't
>need to define such messages in LMP. 
>Thanks 
>Carmine 
>Jonathan Lang wrote: 
>Carmine,  Please see inline.Thanks,Jonathan
>-----Original Message-----From: Carmine Daloia
>[mailto:[e-mail]]Sent: Monday, November 19, 2001 6:44 AMTo:
>[e-mail]: [e-mail]; [e-mail]: LMP:
>Suppression of Downstream Alarms...Hi all,As I read through Section 6 "Fault
>Management", one issue that it seems to be addressing is "Suppression of
>Downstream Alarms".In section 6.2, it states that "If data links fail
>between two PXCs, the power monitoring system in all of the downstream nodes
>may detect LOL and indicate a failure. To avoid multiple alarms stemming
>from the same failure, LMP
> provides a failure notification through the ChannelStatus message...".I
>agree that the suppression of downstream alarms is an important issue.
>great!
>If we look at standard networks (both SONET/SDH and OTN), this capability is
>already provided by the overhead in SDH/SONET and G.709 OTN. G.709 OTN
>handles suppression of alarms in both all-optical networks as well as opaque
>networks. I don't think we need to burden the control plane with such
>functions when the transport plane handles this in standard networks. In
>fact the transport plane handles suppression of alarms on all equipment in
>the network (not just cross-connects).If we look at a pre-OTN
>("non-standard") scenario consisting of Cross-connects, Optical Line
>Systems, and Optical Amplifiers supporting a DWDM networked solution, we can
>analyze two scenarios. One scenario is an opaque network (e.g., the OLS
>supports 3R). In this scenario, the downstream Cross-connects would not
>detect LOL upon faults occurring upstr
>eam. The 3R points on the OLS Line Systems would insert some type of signal
>downstream. Therefore the mechanism described in Section 6.2 does not apply.
>Another scenario is an all-optical pre-OTN network. Note that other
>equipment besides Cross-connects (e.g., Optical Amplifiers) in an
>all-optical network may alarm due to upstream faults. These alarms also need
>to be suppressed. LMP seems to only address the suppression of downstream
>alarms on cross-connects without taking into consideration the network that
>sits between the cross-connects. Is LMP also expected to have to be
>processed on Optical Amplifiers? This seems to be undesirable, especially
>given all the various applications that seem to be included into the LMP
>protocol that would not have anything to do with Optical Amplifieris.
>For interaction between cross-connects and Line Systems, please see
>OLIRequirements
>document(http://search.ietf.org/internet-drafts/draft-many-oli-reqts-00.txt)
>andcorresponding LMP-WDM protocol document (new version to be
>uploadedtomorrow, but old version can be found
>athttp://search.ietf.org/internet-drafts/draft-fredette-lmp-wdm-02.txt).
>Any other views?Carmine
>