The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2003-Oct> msg00155



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

suggestion for draft-ietf-mpls-lsr-mib

  • From: "Thomas D. Nadeau" <tnadeau@cisco.com>
  • Date: Mon, 27 Oct 2003 09:30:28 -0500
  • Importance: Normal
  • Organization: Cisco Systems, inc.
  • X-MIME-Autoconverted: from quoted-printable to 8bit by cell.onecall.net id h9REoEZ18123


	Hi,

>-----Original Message-----
>From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET] On Behalf 
>Of Stefan Winter
>Sent: Sunday, August 10, 2003 9:32 AM
>To: mpls@UU.NET
>Subject: suggestion for draft-ietf-mpls-lsr-mib
>
>
>Hello!
>
>During implementing a management agent for the MPLS-LSR-STD-MIB a minor
>missing feature in the MIB showed up (at least in my understanding).
>The issue is about InSegments. When generating a new 
>InSegment, one must
>specify the type of label, "generic","atm" or "frame relay". 
>The type of label cannot be directly read. 

	I don't understand. You know the interface on which
the label is being allocated, and hense the label space
that it is being allocated from. This tells you whether it
is the per-platform or per-interface space. In any event,
the interface tells you where the label came from.

>The current way to do so is to 
>lookup the
>mplsInSegmentInterface column of the row, then look into 
>IF-MIB::ifType.x
>and evaluate the ifType to guess which of the three is 
>appropriate. Apart
>from the fact that looking up content of another MIB may be 
>expensive, it
>not at always possible to determine the type of label. If, for example,
>ifType reports that the device is an ATM device, both the "atm" or the
>"generic" label types are possible (since "generic" is always 
>an option).


>To overcome that limitation, two options seem to be possible.
>a) extend the mplsInterfaceTable with a column mplsInterfaceLabelType
>b) extend the mplsInSegmentTable with a column mplsInSegmentLabelType
>
>where b) is a more general solution as it allows to specify the type
>per-label and not only per-interface, as solution a) would do. 
>The Syntax
>for such a column could be INTEGER { gen(0), atm(1), frame-relay(2) }.
>And as I write this, I realize that even OutSegments need to 
>have the type of label specified... so the same limitation applies to 
>OutSegments, too.

	The interface table already contains a variable that
indicates the label space participation type which applies
to all labels (segments) allocated using that interface: 

mplsInterfaceLabelParticipationType OBJECT-TYPE
   SYNTAX  BITS {
                  perPlatform (0),
                  perInterface (1)
              }

	--Tom



>
>These are just my 2 cents from an implementorīs view, a 
>comment would be
>nice.
>
>Greetings,
>
>Stefan Winter
>