Network Layer Protocols

Network Layer Protocols are of two types:

1. Connection Oriented Protocols Eg. X.25 2. Connectionless Protocols Eg. IP

Network Layer Protocols (cotd.)

Connection Oriented Protocols: Employ virtual circuits Routing is decided at the VC setup time Provides End-to-End Flow and Error Control Delivers all packets in proper order Less overhead in packets Loss of node/VC causes loss of all packets in VC Unable to adapt to network traffic conditions

Network Layer Protocols (contd.)

Connectionless Protocols No end to end connectivity No flow control or error control by network layer. (It is left to the Transport layer.) Each packet is independent and each finds its route. Packet header must carry full information. Packets can arrive at the destination out of order. No circuit set up time involved. More robust. Can adapt to network changing network traffic conditions. Loss of a node causes only the loss of packets there.

Network Layer Protocols (contd.)

X.25 Protocol
Connection Oriented Protocol ITU-T Recommendation Introduced in 1976. Defined ad the interface between DTE and DCE. i.e. the interface between Packet Mode Terminal and Packet Switching Exchange. DTE Data Terminal Equipment DCE Data Circuit-terminating Equipment Employs Virtual Circuits (either Switched Virtual Circuit (SVC) or Permanent Virtual Circuit (PVC)).

X.25 (contd.)
X.25 Layers Physical layer X.21 or X.21 bis (EIA RS232 equivalent) Link Layer LAP B (HDLC in ABM) Network Layer X.25 LAP B I frames to send data S frames for flow and error control U frames to setup and manage connection Eg. SABM followed by UA followed by several I frames and then a DISC and UA to terminate.

X.25 (contd.)
X.25 Packet Layer Called the Packet Layer Protocol (PLP) Responsible for (a) Establishing connection, (b) Transferring data, and (c) Terminating connection. Responsible for creating the virtual circuit and negotiating service options between DTEs. Note: Frame layer (i.e. link layer) is responsible for making connection between DTE and DCE. Packet layer (X.25) is responsible for making connection between Two DTEs. (i.e. end-to-end connection.)

X.25 (Contd.)
X.25 Flow and Error Control Done at both frame and network layers Frame layer does it across a link Network layer does it end-to-end Note: X.25 Virtual Circuit is created at the network layer and not at the link layer. (In contrast, the Frame Relay and ATM are created at the link layer.) X.25 Packet Format

X.25 (Contd.)
X.25 Packet Format General Format Identifier Identifies packet parameters i.e., data or control information, what kind of window flow control, whether delivery confirmation is required. Logical Channel Identifier Identifies the virtual circuit across the local DTE/DCE interface. Packet Type Identifier Identifies the packet type if it is a control packet. User Data This field carries the user data.

X.25 (Contd.)
X.25 Disadvantages Was designed for PSTN which was slow at the time. Max. speed available is 48 kbps. This is insufficient for modern day data transactions. Addressing scheme (X.21) is based on the telephone numbering system. Has superfluous error control that makes the throughput small. Some X.25 provisions (such as the operation of Character Mode Terminals) have become obsolete.

X.25 Related Other Protocols

X.121 Provides the global addressing scheme used in X.25 packet addressing. Address Format: DNIC (4 digits) NTN (10 digits) DNIC Data Network Identification Code defines the specific network. First 3 digits define the country and the 4th defines the network inside that country. NTN National Terminal Number defines the DTE inside the particular network.

Frame Relay Networks

Frame Relay Features and Advantages Operates at the link layer Virtual circuit mode operation No flow control or error control
Frame error check done only to see if the frame should be passed on or dropped.

Less overhead compared to X.25 in frame format Relies on improved transmission media, such as fiber It particularly suits for protocols like TCP/IP since there is no duplication at the network layer

Frame Relay Networks (Contd.)

Frame relay doesnt need fixed bit rate (like X.25) and thus is more suitable for bursty data. Frame relay frame size can be up to 9000 bytes and thus can accommodate all LAN frames. Frame Relay Disadvantages Frame relay can operate at bit rates up to 44.376 Mbps. But this is not enough for some services. Frame relay allows variable frame lengths. This can create unfair delays for some users. Because of the variable delay (due to variable frame length) frame relay is not suitable for real time traffic.

Frame Relay Operation

FR provides both switched and permanent VCs FR provides connection of LANs or Main Frame Computers to each other in a WAN. The facility that provides the connection on the LAN or Main Frame side is the DTE, and could be a Router, a Bridge, or a Man Frame Computer. The facility that provides switching of Frames in the WAN is the DCE, and is a switching centre. Since Frame relay operates with VCs, the packets carry VC identifiers and not addresses. This identifier is called the Data Link Connection Identifier (DLCI).

Frame Relay Layers

Physical Layer No specific protocol is specified for the physical layer of the Frame Relay. Frame relay supports any of the protocols recognized by ANSI. Data Link Layer At the data link layer FR employs a simplified version of HDLC called LAP F. Flag Address Information FCS Flag Note that the Control field is missing The Address field is divided into several subfields:

Frame Relay Data Link Layer (contd.)

Address Field Format: DLCI C/R EA
6 bits 1 bit 1 bit

DLCI
4 bits

FECN BECN DE EA
1 bit 1 bit 1 bit 1 bit

DLCI (6+4=10 bits) gives the VC identifier. Command/Response bit is for the use of the upper layer. EA is Extended Address bit. If it is 0 then the address field has another byte to follow. If it is 1 the current byte is the last. FECN is Forward Explicit Congestion Notification. BECN is Backward Explicit Congestion Notification. DE is Discard Eligibility bit. If 1 do not discard.

Congestion Control in Frame Relay

Congestion occurs when inflow data at a node exceeds its handling capacity. Congestion can bring down a network quite rapidly. X.25 has flow control at both link and network layers. At link level it controls the flow of data across a link. At network layer it provides end-to-end flow control. Frame Relay has no flow control. Frame Relay allows the user to send bursty data. This makes Frame Relay networks vulnerable for traffic congestion. Therefore some congestion control mechanism at frame level is required. Two bits, BECN and FECN, do this.

Congestion Control in Frame Relay (Contd.)

BECN warns the sender about the congestion in the network. This warning can come (from the receiver) as a separate (special) frame, or in another receiving frame if it is a full-duplex link. FECN warns the receiver that congestion has occurred in the network. Frame relay it self cannot do anything about it but it can communicate with the sender and seek assistance from higher layer protocols. When senders are not responding to congestion notifications the Frame Relay switches will discard frames, depending on the status of the Discard Eligibility bit. Senders are informed by the upper layers (such as Transport layer which will recognize the congestion) about the discarded frames, and will be requested to resend those discarded frames.

Frame Relay Congestion Control

Leaky Bucket Algorithm Behavior of Frame Relay Switch is similar to a leaky bucket that has a small hole at the bottom. The water leaks from the bucket at a constant rate irrespective of the inflow to the bucket (as long as the bucket is not empty). If the inflow to bucket is too rapid the bucket may overflow. Each Frame Relay Switch can be set to send data at a certain rate. If data is received at a rate faster than the sent out rate the switch is congested and the data frames will have to be discarded.

Frame Relay Congestion Control

Ex. A one inlet one outlet frame relay switch with buffer has an input data stream that comes in bursts at a rate of 40 Mbps. The burst duration is 100 ms. The gaps between data bursts are 1 min. long. If the switch has to avoid congestion, what should be the minimum buffer size? Minimum buffer size = 40 Mbps x 100 ms = 4 Mb Ex. If the output line has a capacity of 1.544 Mbps can congestion be avoided with this buffer? Time taken to empty the buffer = 4/1.544 sec. Since this is less than 1 min. the congestion can be avoided.

Frame Relay Traffic Control

Congestion strategies require Frame Relay to take Traffic Control measures to determine when to set BECN, FECN, DE, etc. and to discard frames. For this four different traffic attributes are taken into account. Access rate Committed Burst Size Committed Information Rate Excess burst rate These are set during the VC establishment phase after negotiations between the user and the network.

Frame Relay Traffic Control (contd.)

Access Rate For each connection the access rate (bps) is defined. The value depends on the bandwidth of the channel connecting the user to the network. User cannot exceed this rate. Eg.If the access is through a G703 link, then the access rate cannot exceed 2 Mbps. Committed Burst Size This is the maximum number of bits the network is committed to transfer in a predefined period, without discarding any frames (or setting DE bit). Eg. 400 kb in 4 sec. (could be 300 kb in 1st sec. +100 kb in last sec.)

Frame Relay Traffic Control (contd.)

Committed Information Rate This is the average rate of information the user can send to the network. As long as this is not exceeded the frames will not be discarded. The value of Committed Information Rate (CIR) is related to Committed Burst Size (CBS). CIR = CBS/Burst Time Excess Burst Size This is the maximum number of bits in excess of CBS that a user is allowed to send in a predefined period. Network will honor this if there is no congestion.