ATM BASICS - Chapter 6 pptx

Despite the fact that some ISPs and data-only carriers treat ATM as a data transport solution, ATM is actively used to provision voice and TDM ser-vices, and handles a significant amount

Where is ATM Used Today

This chapter presents the most commonly used ATM applications It

identi-fies market segments as well as operators using ATM infrastructure The

chapter provides basic information referring to the transmission of IP

pack-et over ATM npack-etworks It also covers some of the ATM and Frame Relay

interworking aspects Finally, the reader is given the short introduction to

different solutions for transmission of voice services using ATM

infrastruc-ture

Over last ten years a number of new technologies have become

commercial-ly available The developments in transmission technologies such as the

ability to transmit data at different wavelengths (DWDM) combined with

the introduction of techniques such as Gigabit and 10Gbit Ethernet changed

the market situation Additionally, the advent of MPLS and advances in

QoS based routing in IP network highly influenced the position of ATM

In the public network segment, ATM is important for services delivery,

ser-vice aggregation and transport It is a well-defined worldwide standard that

has been widely deployed because it works and it is not only for data This

technology is embedded into carrier edge and backbone infrastructures to

support enterprise-class applications with high reliability In fact reliability

is the key ATM advantage when compared to all–IP based solutions

Despite the fact that some ISPs and data-only carriers treat ATM as a data

transport solution, ATM is actively used to provision voice and TDM

ser-vices, and handles a significant amount of frame relay and IP services

traf-fic It is also the primary technology used for DSL traffic aggregation and is

used to feed traffic into other service backbones like PON (Passive Optical Network) On the other hand ATM has clearly received a more favorable reception in service provider networks, where its first mission was to pro-vide a scalable core for fast-growing frame relay networks Large carriers are looking for solutions that let them continue to deploy proven and legacy services generating most revenue, e.g voice, frame relay and private line In fact, the capital markets are driving them to this conclusion At the same time, there are ongoing efforts to prepare the networks for an eventual transition to an IP core

In addition it is important to note that ATM makes a lot of sense for multi-service delivery in access networks, since it has the ability to deliver voice via its CBR service, as well as data via VBR and UBR ATM remains the most cost-effective method for transporting real-time traffic at OC-12 rates and below This is proven by the number of wireline and wireless voice car-riers who use ATM to transport their voice traffic

Concluding, it is possible to identify a few applications that currently domi-nate as far as the usage of ATM is discussed The first and probably the one that would decline in next few years is the transmission of IP packet over ATM (IP over TM) The second is ATM and Frame Relay interworking used

at large scale by carriers offering Frame Relay services Last but no least, the growing popularity Voice over ATM solutions would be also covered There are also many other ATM applications but their market penetration

is today relatively small and either they have already found strong com-petitors or they would be replaced soon by solutions mostly based on IP plat-forms They include mainly applications using SVC services such as: multi-media, Video Dial Tone and Conferencing Service, Interworking with exist-ing LANs (LANE), Campus/Corporate Enterprise Networks (MPOA)

6.1 IP over ATM

The initial success of ATM lied largely in its ability to transport legacy data

traffic, mostly IP, over its network infrastructure However due to number

of different factors ATM has had a decidedly mixed history in the ISP world

Looking back to the early 1990s, ISPs’ networks consisted of routers

inter-connected by leased lines However, with the growth rate of the Internet ISP

network operators were forced to migrate to higher-speed technologies by

the mid-1990s At that time, ATM was available at the higher speed of 155

Mbps, and soon after at 622 Mbps High-capacity ATM switches were also

significantly less expensive than high-capacity IP routers Consequently,

ATM became the backbone technology of choice for most of the world’s large

ISPs

Today ATM has lost quite a lot of its attractiveness on behalf of IP and

MPLS-based solutions ATM technology as a way to implement network

backbones, is not all that important in the ISP world any more A few ISPs

do use ATM to do traffic engineering (laying down specific paths for traffic

aggregates to avoid overloaded network links) A few others use ATM

because it can be a cost-effective layer-2 interconnect or because ATM is the

only high speed service offering they can get in some locations But the fact

that the underlying technology is ATM layer is only currently relevant to

those ISPs that are performing traffic engineering, and with the rollout of

MPLS, ATM will become less required Deeper in the core ATM offers little

benefit MPLS could completely replace ATM in core networking

applica-tions, providing that the MPLS players do a better job of building LSRs that

are box-for-box evolutions of ATM switches Here it is important to note that

ATM switches can be upgraded to MPLS routers In fact building MPLS

net-works using ATM infrastructure is widely concerned as one of the most

important methods for quick and efficient MPLS deployment What is also

important, the ATM and MPLS can coexist in the same switching device

This case is referred to as ‘Ships In the Night’

Finally, ATM is an important technology in some access networks,

especial-ly in the access portion of the network, where bandwidth and QOS are crit-ical This particularly refers to DSL Yet the fact that it is ATM is

general-ly irrelevant—all that is needed is a way to provide virtual circuits to sepa-rate customer connections in order to ensure QoS

Concluding, the ISPs gradually change the position of ATM in their net-works They consider ATM as the technology suitable for the access part of their networks rather the technology for the core network, where ATM is being rapidly replaced with MPLS

Fig 6-1, LLC/SNAP Encapsulation

Over a decade a number of different methods for IP and ATM integration

have proposed, standardized and implemented Some of them are still in use

but most of them appeared to be unsuccessful (e.g LANE, MPOA)

Network layer packets can be transported over ATM network in two

differ-ent ways Firstly, it is possible to transmit differdiffer-ent protocols over the same

VC The protocols are distinguished with the help of additional headers

This method is called LLC/ SNAP encapsulation and it was defined by the

IETF (initially in RFC 1483) Secondly, each protocol is carried over a

sepa-rate virtual channel This method is called VC-base multiplexing

The LLC header is placed in front of the PDU that is carried in the payload

field of the CS of AAL5 The three-octet LLC header contains information

that identifies the protocol of the PDU If a routed ISO protocol is

encapsu-lated, the

PDU follows the header directly and the protocol is identified in the

proto-col

data, using a Network Layer Protocol Identifier (NLPID) When a routed

non-ISO protocol is encapsulated, the LLC header is followed by the

Subnetwork Attachement Point (SNAP) header The presence of the SNAP

header is indicated in the LLC header

The SNAP header contains two fields (the structure of the SNAP header is

shown in Fig 5.7): the 3-byte Organizationally Unique Identifier (OUI) and

the 2-byte Protocol Identifier (PID) A combination of OUI and PID values

indicates the particular bridged or routed protocol, such as IPv4 for

instance

The simplest and the most popular method for IP and ATM interworking is

Classical IP over ATM (CLIP), an IETF standard for internetworking IP and

ATM, firstly described in IETF RFC 1577 Currently CLIP is defined in RFC

2225 CLIP was the first method of internetworking IP and ATM, and was

developed at a time when ATM technology was immature CLIP deploys an

‘overlay model’, which means, that CLIP effectively ignores ATM’s

proper-ties, treating it as a transmission technology or as simply underlying ‘wires’

used to carry IP packets CLIP preserves the classical model of IP routing,

that is, the end-to-end IP routing architecture stays the same Simply

stat-ed, this means that traffic wanting to go from one IP subnet to another IP

subnet has to go through an IP router with all the consequences CLIP oper-ates in much the same way as IP over Ethernet In the case of CLIP, it is necessary to map destination IP addresses to destination ATM addresses in case of SVC connections, or in a PVC environment, to map IP addresses directly to VPI/VCI values In an Ethernet network the task of mapping IP addresses to hardware addresses is assigned to ARP (Address Resolution Protocol), which extensively makes the use of Ethernet broadcasting capa-bilities In CLIP model a modified version of ARP, called ATMARP, is used

to accomplish address mapping The CLIP architecture varies depending on the type of virtual connections

In SVC environment (shown in the Fig 6-2) an ATMARP server must used

as the mechanism for hosts to resolve destination IP addresses to destina-tion ATM addresses An ATMARP server provides its services within the Logical IP Subnet (LIS) The LIS is used to refer to an IP subnet deployed over an ATM network as part of CLIP Any transfer of IP packets directed

to a host outside the LIS must involve an IP router This is a must even though the two hosts can be the members of the same ATM network In CLIP model a host that wishes to send data to another host must register its network and ATM addresses in the ATM ARP server Then it can issue

a query to the server to obtain the destination ATM address for a given IP address Once the server resolves and IP address into ATM address, a host can establish a direct SVC to the destination host

CLIP model in a PVC environment, which is more popular nowadays,

assumes that the operation of mapping between destination IP addresses

and VPI/VCI values can be done without the use of any server However,

ATM ARP messages are used to check the destination ATM addresses for a

host on its open virtual connections Additionally, manual dimensioning of

PVCs can be treated as the rudimentary mechanism for traffic engineering

especially for networks of well-defined and stable architecture This

approach is used, for instance, by GSM/GPRS operators within their IP

backbone networks provided ATM is the underlying technology

Fig 6-2, Classical IP over ATM in SVC environment

6.2 Voice over ATM

The major source of revenue for most operators around the world is tele-phony services Voice services generate over $200 billion in U.S revenues and it is likely to remain so for the coming years – while the main source of growth will be IP-based services This means that networks have to be opti-mized for carrying packet-oriented traffic, but at the same time they must deliver a reliable and high quality voice service Voice is considered by most

to be a commodity, and ATM has played a role in lowering the cost of ser-vice for some carriers Although most recent innovations in voice serser-vices are based on IP and MPLS, and end-to-end voice over IP architectures promise far greater cost reductions than result from simply using ATM for bulk transport The ATM is continuously considered as the mature and reli-able technology This point of view is especially typical to large incumbent telecom operators who were involved in the ATM standardization process They are therefore generally in favor of this technology, since it offers a safe migration path for them

6.2.1 Circuit Emulation Services

Circuit Emulation (CE) represents today a stable and reliable standard, which has been widely implemented by ATM equipment suppliers The CE function enables existing TDM circuits to be mapped over ATM CE thus give the chance to migrate an existing TDM network to ATM whilst pre-serving the previous investment in TDM equipment Circuit Emulation products are available for all major circuits, including the American T1 and European E1 standards When using CE the ATM network simply provides

a transparent transport mechanism for structured TDM circuits Voice is encoded into these links as in a normal TDM network (e.g PSTN) using PCM, ADPCM, or other encoding & compression mechanisms The network will ensure that the delivered circuit is reconstructed exactly as received Circuit Emulation uses the AAL1 mechanism to segment the incoming E1

or T1 traffic into ATM cells with the necessary timing information to ensure that the circuit can be correctly reassembled at the destination The Fig 6-3 shows the typical CE application

In the figure an incoming E1 circuit is plugged into the ATM switch The

ATM switch then performs the CE function and transmit cell using the CBR

VCC The interworking between TDM and ATM standards may be achieved

in a number of ways It may be the case that the ATM switch has a special

CE module, which is a plug in board with E1 or E1 interfaces and

Application Specific Integrated Circuits (ASICs) to perform the CE function

Furthermore, the ATM switch may have an E1 or T1 interface cards and the

CE function may be a software function running on the main switch

proces-sor It is also feasible to for the PBX (or similar TDM equipment) to have an

ATM interface board In this case the voice samples are sent out within

ATM cells from the PBX towards the ATM switch

Fig 6-3, Architecture for CES

CE can be deployed in two different modes: structured and unstructured In unstructured CE mode the network does not attempt to recognise the inter-nal circuit structure Rather it simply transmits the entire circuit across the network At the destination the original stream of bits is reconstructed This method doesn’t allow for accessing particular TDM voice channels repre-sented by time slots within ATM network Hence, the unstructured CE car-ries the entire circuit across the ATM network with no recognition of the internal framing structure This unstructured service can emulate for instance a 2 Mbps data leased line

In structured CE mode the ATM network recognises the internal structure

of the circuit and is able to recover this structure at the receiving end Circuit structure refers to the timeslots E1 for example, contains 32-times-lot frames, whereas T1 contains 24 times32-times-lots An example of structured CE

is given in the Fig 6-2 In this example single timeslots are extracted from the source E1 and mapped to an ATM cell This means that particular timeslots may be mapped to different virtual circuits and consecutively to different destinations Several timeslots from a source circuit may be mapped to one virtual circuit It can be easily seen in the example that all time slots no 1 from each frame is mapped to an ATM cell Therefore, depending on the TDM circuit type certain latency is inevitably introduced

In the worst case 47 time slots representing a voice connection must be col-lected before the cell can be created This can highly influence QoS observed

by the applications In order to overcome this limitation a partial fill mech-anism can be used, which means that cell recognised as a long distance or international call can be heavily padded Another option can be to map sev-eral timeslots of different voice connections into one cell and therefore one

VC