2. Credit Window Flow Control

This page preserves the original RFC text for the section and adds navigation metadata for the documentation site.

Original RFC Text

2.  Credit Window Flow Control

   This section defines additions to DLEP used in credit-based flow
   control.  The additions provide the DLEP mechanisms to control
   credits.  Routers then use this information to regulate when data is
   sent to a modem.

   The credit window flow control mechanisms defined in this document
   support credit-based flow control of traffic sent from a router to a
   modem.  The mapping of specific flows to a particular credit window
   is based on the Traffic Classification Data Item defined in
   [RFC9892].  Both types of DLEP peers -- router and modem -- negotiate
   the use of an extension employing this mechanism during session
   initialization as required; for example, see [RFC9894].  When using
   credit windows, data traffic is only allowed to be sent by the router
   to the modem when there are credits available.

   Credits are managed on a 'per logical "credit window"' basis.  Each
   credit window can be thought of as corresponding to a queue within a
   modem.  Credit windows may be shared across, or dedicated to,
   destinations and data plane identifiers -- for example, DSCPs -- at a
   granularity that is appropriate for a modem's implementation and its
   attached transmission technology.  As specified in Section 2.3.1,
   there is a direct one-to-one mapping of credit windows to flows as
   identified by Flow Identifiers (FIDs) carried within the Traffic
   Classification Data Item.  Modems pass to the router information on
   their credit windows and FIDs prior to a router being able to send
   data when an extension requiring the use of credit window flow
   control is used.  Note that Traffic Classification Identifier (TID)
   values and FID values are significant only to the issuing modem.
   There is no relationship implied by the same TID or FID value being
   issued by more than one modem.  In addition to the traffic
   classification information associated with a FID, modems provide an
   initial credit window size, as well as the maximum size of the
   logical queue associated with each credit window.  The maximum size
   is included for informative and potential future uses.

   Modems provide an initial credit window size at the time of credit
   window initialization.  Such initialization can take place during
   session initiation or any point thereafter.  It can also take place
   when rate information changes.  An increment to a credit window size,
   specified in a Credit Window Grant Data Item, is provided in a
   Destination Up Message (Section 2.3.2) or Credit Control Message
   (Section 2.2.1).  A router provides its view of the credit window,
   which is known as "Status", in Destination Up Response Messages
   (Section 2.3.3) and Credit Control Response Messages (Section 2.2.2).
   Routers can also request credits using the Credit Control Message.

   When modems provide credits to a router, they will need to take into
   account any overhead of their attached transmission technology and
   map it into the credit semantics defined in this document.  In
   particular, the credit window is defined below to include per-frame
   (per-packet) Media Access Control (MAC) headers, and this may not
   match the actual overhead of the modems' attached transmission
   technology.  In that case, a direct mapping or an approximation will
   need to be made by the modem to provide appropriate credit values.

   Actual flows of traffic are mapped to credit windows based on flow
   identification information provided by modems in the Traffic
   Classification Data Item defined in [RFC9892].  This Data Item
   supports traffic classification on a per-destination or more fine-
   grained level.  Routers use the combination of the DLEP-identified
   destination and flow information associated with a credit window in
   order to match traffic they send to specific credit windows.  In some
   cases, the Traffic Classification Data Item allows the modem to
   specify a wildcard to match any packets that do not match other Data
   Items; for example, see [RFC9895].  In the absence of a wildcard, a
   packet may not match any of the Data Items and, in this case, MUST be
   dropped by the router.

   When a destination becomes reachable, a modem "associates"
   (identifies) the appropriate traffic classification information via
   the TID to be used for traffic sent by the router to that
   destination.  This is supported by the Credit Window Association Data
   Item, which is carried in Destination Up and Destination Update
   Messages; see Section 2.3.2.  The TID provides the information to
   support router traffic classification, based on the FIDs contained in
   the TID; see [RFC9892].  As defined, each credit window has a
   corresponding FID, so traffic is mapped to a credit window by
   locating a matching FID that is contained in the TID that is
   associated with the traffic's destination.  This means that the use
   of FIDs and TIDs, and the association of a TID to a DLEP destination,
   enable a modem to share or dedicate resources as needed to match the
   specifics of its implementation and its attached transmission
   technology.

   Credit window flow control as defined in this document has objectives
   similar to the control technique described in
   [Credit-Window-Extension].  One notable difference from that type of
   credit control is that in this document, credits are never provided
   by the router to the modem.