Frame Relay

Document Sample
Frame Relay Powered By Docstoc
					                        Frame Relay
• Frame Relay is a connection-oriented data transport service for public
  switched networks.
• The Frame Relay protocols are a modification of X.25 standards
  which came in 1994. X.25 was able to satisfactory perform with noisy
  transmission line.
• Both X.25 and Frame Relay specify the lowest three OSI layers as
  shown in the next page
• The interface from the user to the frame-relay network is called
  frame-relay interface (FRI). FRI currently supports the following
  access speeds:
    – 56 kbps
    – n x 64 kbps
    – 1.544 Mbps (T1) and 2.048 Mbps (El for Europe)
    – There are also some intermediate speeds, but the highest is T1 or

        1                         2                 ...        k                p = 1 Ð PER; q = pk

                                                                        k2/q < k1/p when p Ð 1
Link error control: k1/p
                                               Time w/o error control             Time with error control
  error control: k2/q

                                              Frame structure

 Flag           Header        User data field (integral number of octets)           FCS       Flag

  1               2                                 Variable                         2          1    bytes

                                                                         DLCI = Data link connection
                      DLCI                                         EA      identifer
                                                                         F/BECN = Forward/backward
                      DLCI                     DE                          explicit congestion notification
                                                                         C/R = Command/response
            7                 4       3   2     1     0        bits      DE = Discard eligibility
                             FECN         BECN                           EA = Extended address

                   Frame Processing Speed
•    FR is an enhanced version of X.25 without error correction at the
     intermediate nodes. Error corrections in a FR network occur at either ends
     of the connection, thus making frame relay much faster than X.25.
•    Also, the transmissions on a link are not slowed down as they use “Go Back
     N” protocol of X.25. Thus, FR Relay is a virtual circuit service which does
     not provide reliability.
•    Prove that the end-to-end delay of FR << End-to-end delay for X.25
    1.   Let, FER between two nodes = 1-p
    2.   Thus, P{No frame error between nodes}= p
    3.   Let, the total number of nodes through which the frame (packet) passes = k
    4.   Thus, E[ Number of transmissions before receiving a correct frame ] = 1/p
    5.   Let, 1be the total delay at the node for X.25 (Processing delay + Transmission
         delay within the switch= 47 msec + 3 msec = 50 msec; suggested in the book)
    6.   Total end-to-end delay for X.25 = k 1/p
    7.    For FR delay at the node = 2 = Transmission delay at node = 3 msec (in this
    8.   Total end-to-end delay = k 2/pk;
                    where, pk = P[Correct frame being received by the end node]
          If, P[correct frame received by adjacent nodes] 1, Then pk  p
    9.   Therefore, k 1/p >> k 2/pk or, FR delay is lot less than X.25 delay

                       Frame Format
• Frame is shown in Figure above. Explanation for different
  control fields follows:
   – Total control bytes (Header) can either be 2 or 4 designated by
     EA(Address extension bit). EA = 0 next byte is also an address
   – DLCI +EA (10 +2 ) bits used for routing the frame
   – DLCI (Data Link Connection Identifier) = 10 bits. Can address up to
     1024 PVCs. Out of a total of 1024, one thousand are assigned
     between two users when they subscribe for FR service per access
     link (Figure). At the network node this connection specification is
     confirmed. If the specification is in error the frame is discarded or
     else the frame is relayed to its destination (See attached figures for
     PVC and virtual connections, and the flow chart stating conditions
     of discarding frames).
   – F/BECN (Forward/Backward explicit congestion notification) + DE
     are used for notification and enforcement as discussed below

                       Frame Format
       • DE: (Discard eligibility) bit may be set by users to indicate
         low-priority frames such as some audio or imaging frames with
         less significant information. On the other hand since
         compressed video is more sensitive to losses and thus such
         packets might not have the DE bit set. DE bit can be set by the
         network for any frame that exceeds a user’s subscribed rate.
         The network assumes that anything that exceeds the users
         subscribed rate is of low priority.
       • F/BECN To reduce buffer overflow, the switch can exercise
         flow control as follows. When a switch experiences some
         congestion, it notifies the source and destination of all the
         active PVCs passing through the node. This is done by setting
         the FECN (forward explicit congestion notification) bit in user
         frames going in the forward direction to inform the destination
         or the BECN backward explicit congestion notification) bit in
         user frames going in the reverse direction to inform the source
         of the overflow (congestion)
• C/R: (Command/Response) Application specific usage
      Congestion control mechanism
• At subscription time, each DLCI is assigned three
  parameters (Tc, Bc, Be) for traffic shaping. These
  parameters are used as follows:
   – Time is slotted into intervals of duration Tc (Burst
     duration). The network guarantees transport of Bc bytes
     of data in each interval. This guarantees a ”Committed
     information rate" CIR = Bc/Tc. If the user injects more
     than Bc bytes across the user-network interface in an
     interval, the network may admit the first Be bytes of
     excess data with their DE bits set. Further frames in that
     interval may be discarded. The DLCI is guaranteed a
     long-term bandwidth of CIR and a maximum burst size
     of Be. This traffic-shaping scheme regulates the input
     load to the frame relay network


Shared By: