Computer network technologies and services/WAN

Strictly speaking, a Wide Area Network (WAN) is a network that is extended over a broad area, spanning regions, countries or in the case of the Internet even the world. More generally, any computer networking technology used to transmit data over long distances can be called as a WAN.

A WAN technology should meet some requirements in terms of service duration, bit rate and delay constraints according to the application (telemetry, telephony, data transfer, etc.) it is designed for.

Asynchronous Transfer Mode (ATM) represents the convergence for a wide variety of technologies that in the past both telecom and IT worlds in parallel introduced in order to transmit data over long distances:

• in the telecom world, the telephony turned from analog to digital, then ISDN and B-ISDN started to carry data along with the voice;
• in the IT world, Frame Relay superseded analog and digital leased lines by taking advantage of packet switching, and X.25 by moving the complexity from core to edge nodes.

Nowadays ATM is going to be abandoned in favour of IP thanks to its lower complexity and greater simplicity.

ISDN

Integrated Service Digital Network (ISDN) allows to carry data along with the voice: a variety of digital devices can be connected to a bus and can transmit over the available ISDN channels:

• Basic Rate Access (BRA) or Basic Rate Interface (BRI): it offers 2 data B-channels at 64 kbps and 1 signaling D-channel at 16 kbps → total: 144 kbps (good for single users or small offices);
• Primary Rate Access (PRA) or Primary Rate Interface (PRI): it offers 30 data B-channels at 64 kbps and 1 signaling D-channel at 16 kbps → total: 2 Mbps (good for companies).

The transmission is based on Time Division Multiplexing (TDM); all channels go to a Network Termination and enter the network over a digital wire called 'local loop'. The channels inherit the logics from telecom operators: they keep being alive also when no data is exchanged.

PDH

PDH hierarchy.

Plesiochronous Digital Hierarchy (PDH) is an old standard designed to transfer digital voice channels at 64 Kb/s (PCM) over TDM-based digital telephone networks. The system is called 'plesiochronous' because a tight synchronization between transmitter and receiver is required, even if each device has its own clock.

Data flows are organized in a hierarchical way: channels are aggregated into flows from the lowest layer to the highest one (grooming), and the higher the hierarchical layer, the higher is the bit rate. For example, at layer T1 24 T0-layer channels are put into a single frame one next to another: as the frame has to last 125 µs for all layers, at layer T1 the bit rate will be 24 times higher than the one at layer T0.

SDH

SDH physical and protocol architectures.

Synchronous Digital Hierarchy (SDH), the European equivalent of the international standard SONET, differs from PDH for its higher speeds:

• a single clock exists for the whole system → a synchronization network is required for a tighter synchronization;
• copper cables need to be replaced with optical fibers;
• the flow multiplexing is more complex than PDH, because it is designed to optimize the hardware processing.

The protocol architecture is organized as a layer stack, and each node in the physical network architecture implements them according to its functionality:

• path layer: end-to-end interconnection between two terminals;
• line layer: a path is split into lines by multiplexers;
• section layer: a line is split into sections by repeaters (for long distances);
• photonic layer: the lowest layer for optical fibers.

Each time frame lasts 125 µs and its header includes synchronization information used to combine and separate channels, and OAM (Operation, Administration and Management) information used to detect failures and recover from them.

SDH and PDH represent the transport layer which ATM and Frame Relay operate on.

Frame Relay

Frame Relay is a layer 2 connection-oriented standard to set up permanent virtual circuits over packet-switching networks. Each permanent circuit is identified by a Data Link Connection Identifier (DLCI).

The standard is very flexible: in fact it does not specify the technology at upper layer (ATM, X.25...) used internally in the network.

CIR

The service is guaranteed for the blue user but not for the green one because his burstiness is too high.

The maximum supported bit rate is not enough to describe the performance of a Frame Relay network, because an user may send bits consecutively at the maximum bit rate (wire speed) for a long period of time causing congestion in the network. Therefore the network provider provides also the Committed Information Rate (CIR), that is the maximum number $B_{\text{C}}$ of bits the user can transmit within a certain interval of time $T_{\text{C}}$ so that the service is guaranteed:

${\text{CIR}}=B_{\text{C}}\cdot T_{\text{C}}$

where $B_{\text{C}}$ is called committed burst size:

• low burstiness: the user rarely sends packets → the service is always guaranteed;
• high burstiness: the user keeps sending packets consecutively at wire speed → when he exceeds the committed burst size the service will not be guaranteed anymore.

The user's Data Terminal Equipment (DTE) can stop the transmission when the maximum burstiness is reached.

ATM

Asynchronous Transfer Mode (ATM) is a connection-oriented standard to set up virtual circuits over B-ISDN networks. Each circuit is identified by a Virtual Path Identifier (VPI) and a Virtual Circuit Identifier (VCI), and it can be permanent or dynamically set up through signaling messages.

ATM cells are very small: each ATM cell is 53 bytes long, made up of a 5-bytes-long header, containing the connection identifiers, and a 48-bytes-long payload → low latency and low packetization delays.

ATM networks have a very complex model, derived from a telecom-operator mentality to have the full control of the network and guarantee a high fault tolerance.

AAL 5

When ATM was designed, it was thought to be implemented ubiquitously in the network, also at its edges in the network cards of the user PCs. Nowadays PCs at the edges are implementing only the IP protocol because its implementation is cheaper, and ATM can be found only as transport layer in the core of the network hidden from the user.

ATM Adaptation Layer (AAL) of type 5 is used for Segmentation and Reassembly (SAR):

• Segmentation: IP packets are split into ATM cells;
• Reassembly: ATM cells are combined into IP packets.

AAL makes interaction between IP and ATM complex, because IP addresses should be translated to ATM connection identifiers and vice versa → nowadays the tendency is abandoning the ATM control plane and adopting the MPLS control plane.

Optical networks

In optical networks data are transmitted over electromagnetic waves multiplexed by using WDM, transported via optical fibers and switched by mirror-based optical switching systems.

Wavelength Division Multiplexing (WDM) allows to put multiple optical signals into a single optical fiber → the transmission capacity of fibers is increased:

• Coarse WDM (CWDM): it allows to transmit a lower number of signals with wavelengths well-separated one from each other → cheaper because demultiplexing is easier;
• Dense WDM (DWDM): it allows to transmit a higher number of signals with any wavelength → more expensive because demultiplexing is more complex.

Optical switching is based on mirrors controlled by micro-electro-mechanical systems (MEMS), reflecting electromagnetic signals from an input fiber to an output fiber. Optical switching is very flexible: it exploits physical properties of electromagnetic waves without caring about bits → networks can be upgraded to higher speeds because optical switches keep working independently of the bit rate.

Several types of optical switches exist:

• add/drop multiplexer: it is the simplest optical switch: it can be interposed between two fibers to optically insert (add) signals coming from transmitters into the network, and extract (drop) signals from the network towards the receivers;
• cross-connect: it can connect multiple input fibers to multiple output fibers:
  • fiber cross-connect: all the electromagnetic waves coming from an input fiber are switched to an output fiber;
  • waveband cross-connect: a set of electromagnetic waves with close wavelengths coming from an input fiber is switched to an output fiber;
  • wavelength cross-connect: a set of electromagnetic waves with the same wavelength coming from an input fiber is switched to an output fiber;
• wavelength switch: configuration is dynamic, that is switches can change circuits faster than cross-connects → fault recovering is fast.

Two signals with the same wavelength may be coming from two different input fibers but they may need to be switched to the same output fiber → through the wavelength conversion an optical switch can change the wavelength of a signal to one not still used in the output fiber, in order to keep all signals separated.

Optical switches can be used in the network backbone to interconnect the major access points, by setting up optical paths via optical fibers among the cities in the world. Optical switches can set up optical paths by using signaling and routing protocols such as LDP and RSVP. Optical switches are fault tolerant: when a link breaks, they can reflect the waves along another optical path.

WDM can be deployed as the transport layer on which any layer 2 protocol (SONET, Ethernet...) can operate delimiting the frames.

However the technology for pure optical switching is still in an embryonic stage: nowadays WDM switches are more expensive than packet-switching ones, and they can have few interfaces because the mirror system would be very complex for a lot of interfaces. Moreover optical switching is connection-oriented: when a circuit is set up, the resources keep being allocated even if the circuit is not currently used → optical switching is good for the network backbone where the traffic is quite continuous.

Cheaper solutions try to overcome technological limits by replacing mirrors with an electrical switching matrix: each optical signal is converted to a sequence of bits through an optical-to-electrical (OE) conversion so that it can be switched more easily, then it is converted again into an optical signal. The reconverted signal is regenerated, being able to travel for a longer distance before losing power, but this solution has a lot of disadvantages: the switches consume a lot of power with respect to all-optical switches, and changing the bit rate requires to upgrade the switches.