• Nenhum resultado encontrado

In-house cabling

N/A
N/A
Protected

Academic year: 2023

Share "In-house cabling"

Copied!
27
0
0

Texto

(1)

In-house cabling

Cabling is the IT investment with the longest life time

Structured cabling

One or several wiring closets on every floor

Enough space for routers and switches

Enough fiber in the backbone cabling between them

Cross connections for fibre and copper in the closets

UTP cat. 5 (currently, in the future probably something else) cabling from the closets to the outlets (max. 90 m)

RJ-45 connectors, 4 per office desk

Adequate power outlets to the closets and office rooms

=> A network that can adapt to changing needs:

Data and telephony in the same network

Traditional LANs: Ethernet, Token Ring etc.

Fast LANs: 100 Mb/s - 1 Gb/s Ethernet

ATM 25 – 155 – 622 Mb/s

(2)

Wireless LANs

Wireless LANs (WLANs) appear in too configurations:

infrastructural - wireless transmission replacing cables

ad-hoc - networks formed without prior arrangements

Most WLANs use spread spectrum technologies:

Frequency Hopping Spread Spectrum (FHSS) - uses a

narrow-band carrier changing frequencies in a pre-defined manner known to transmitter and receiver

Direct Sequence Spread Spectrum (DSSS) - generates a redundant bit pattern ("chip" or "chipping code") for each bit transmitted

Spread spectrum technologies use a wider frequency band than narrow-band technologies but are more robust

Other users of the same frequency band hear spread spectrum traffic as background noise

(3)

WLANs

The IEEE 802.11 WLAN standard is already implemented in commercial products

Current products operate at 2.4 GHz frequency at 11 Mb/s

New products at 5 GHz frequency an 54 Mb/s are coming to the market soon

100 Mb/s and faster products already operate in laboratories

Blue tooth offers low-cost low-capacity WLAN technology for all kinds of devices

WLANs promise high capacities at low cost within buildings

Their greatest strength is that they operate at unregulated frequencies

Outside buildings this may also be their greatest weakness

(4)

Network Layer

The main purpose of the network layer is to take data packets through the network from the source to the destination

Network layer normally is the uppermost layer provided by the telco, layers 4 through 7 being end-to-end layers

implemented in the interconnected devices

For a long time there was an almost religious argument between two approaches to network services:

Traditionally network services provided by telcos have been connection-oriented and reliable

The service provided by the Internet is connectionless and unreliable

Tanenbaum is using ATM as an example of the traditional thinking in today's world

In fact telcos have already lost the battle to the Internet

(5)

Functions of the network layer

The central functions of the network layer are:

Routing

Congestion control

Quality of service (QoS)

Tanenbaum treats routing and congestion control at length - make sure to read the book

Quality of Service as a whole is a larger issue than just congestion control

ATM will be treated briefly as an example of another kind of approach to networking

IP routing will be discussed briefly here and in more detail on the course "Tik-110.350 Computer networks"

IP multicasting, mobile IP, DiffServ etc. ditto

(6)

Distance Vector Protocols

Distance vector protocols (or Bellman–Ford protocols) are the oldest type of routing protocols

For example, the distance vector containing distances from A could look like this: (B=1, C=2, D=1, E=2)

Each node calculates its routing table and sends its distance vector to its neighbors

The shortest distances to networks or nodes and the links corresponding to these routes are stored in the routing table and changes are reported to the neighbors

To make sure, the routers send each other distance vectors with certain intervals (even if there are no changes)

Distance vector protocols suffer from many problems, such as slow convergence and bouncing

The worst problems with distance vector protocols are that they don't scale well and cannot guarantee loop-freeness

(7)

RIP Example

In the following example we are observing a small network of RIP routers

In the picture on the following page, all other networks have cost "1" except of the link C-D which has cost "10"

We are looking at the entry corresponding to network N in the routing tables of A, B, C and D

The table on the right of the picture shows distances to N from the routers when the link B-D suddenly goes down (distance grows from 1 to "infinite" = 16)

For simplicity, we assume that all the routers send their

routing updates at the same instant (e.g. with 30 s intervals)

The table at the bottom of the next page shows the evolution of the routing tables with time growing from left to right

(8)

RIP Example

D: directly connected, metric 1 B: route via D, metric 2

A: route via B, metric 3 C: route via A, metric 4

time --->

D: dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 dc,1 B: ur A, 4 A, 6 A, 6 A, 8 A, 8 A,10 A,10 A,12 A,12 A,13 A: B, 3 C, 5 C, 5 C, 7 C, 7 C, 9 C, 9 C,11 C,11 C,12 C,12 C: A, 4 A, 4 A, 6 A, 6 A, 8 A, 8 A,10 A,10 D,11 D,11 D,11 dc = directly connected

ur = unreachable

A B

C

D

1 1

10 1 → ∞∞ ∞∞ (16)

N 1

(9)

Link state protocols

Link state protocols were developed at the ARPANET as a solution to the problems associated with traditional routing protocols

Instead of exchanging distance vectors, each node maintains a topological map of the whole network

The maps are updated fast (using a flooding algorithm) after each change in the network topology

With the help of the complete topological map, each node can calculate the optimal routes to all other nodes

Routes can be computed as precisely as if the computation was centralized (even though it is distributed)

The early work done at the ARPANET has led into a number of new link state protocols, such as the ISO IS-IS and OSPF

The Internet Architecture Board (IAB) recommends OSPF to substitute RIP

(10)

Shortest Path First algorithm

The Shortest Path First (SPF) algorithm of Dijkstra calculates efficiently the shortest distances from a node to other nodes in the network

The algorithm works as follows (s = start node, N = set of all nodes):

1. Initialize: ”examined nodes” E = {s},

”remaining nodes” R = {N-s},

”ordered list of paths” O = {all one-segment paths originating in s}

(the cost of each path equals the cost of the link).

2. If O = or if the cost of the first path in O = , mark all the nodes in R

”unreachable” and terminate the algorithm.

3. Let P be the shortest path in list O, remove P from O.

Let v be the last node in P, if v∈∈∈∈E then continue at step 2 else P is the shortest path from s to v, move v from R to E.

4. Build a set of new paths by concatenating P with each link starting from v. The cost of each path is the cost of P + the cost of the link appended. Add these paths into O so that O remains sorted in

ascending order. Goto step 2.

(11)

Basics of ATM

ATM stands for Asynchronous Transfer Mode and has the following characteristics:

A Cell Switching technique

Working in the data link layer of the OSI model

Connection oriented

The small, fixed-size (53-octet) cells make hardware implementations efficient and economical

ATM can be used on a number of various physical layers =>

it can easily adapt to new transmission techniques and transfer speeds

ATM can be used in WANs, MANs and LANs which should lead into seamless integration of services

ATM WAN infrastructure will mainly be based on SDH (Synchronous Digital Hierarchy) and

SONET (Synchronous Optical Network) infrastructure

(12)

B-ISDN

The development of ATM was started by CCITT (now ITU-T) as the core technology of the future Broadband ISDN (B-ISDN)

The idea was, that the PSTN would evolve as follows:

Narrowband ISDN would be implemented by digitalizing the local loops of a digital telephone network

Broadband ISDN would be the next step including e.g:

Distribution of digital audio and video

A customer-premises broadband LAN

The standardization process at CCITT was slow

ATM Forum (ATMF) was formed by American manufacturers of communications equipment to speed things up

ATMF aimed at interoperable working solutions fast

This lead into two parallel standardization processes

ATM is today mainly used in LAN-interconnection

(13)

Virtual Circuits and Virtual Paths

ATM is connection-oriented

The basic abstraction of ATM is the Virtual Channel (VC)

which can be unidirectional and

whose id only has local significance

A VC is contained within a Virtual Path (VP)

which is a bi-directional point-to-point connection between two ATM switching points and

only has local significance

Each VC link is identified by its VC identifier (VCI) included in the ATM cell header

Each VP link can contain a number of VCs and is identified by its VP identifier (VPI) also included in the ATM cell header

A VC Connection (VCC) is a sequence of VC links connected at ATM switches

An ATM switch can switch VCs or entire VPs

(14)

Cell Switching

The PSTN is based on time division multiplexing (TDM) and circuits of fixed capacities

In TDM certain time-slots are assigned for each connection, that means the communication is totally synchronous

In traditional telco networks frames are repeated at 125 µµµµs

intervals corresponding to the 8 kHz sampling rate of speech

The other extreme is datagram services with best-effort

delivery and no bandwidth guarantees, such as LANs and IP networks which are totally asynchronous

In virtual connection based packet switched networks, band- width and some delay guarantees can be given to connections

Cells are relatively short, fixed-size packets, which can be sent in an isochronous manner

Cell switching makes it possible to offer real-time and non-real-time services on the same network

(15)

Cell Switching

In TDM, each time-slot is assigned to a certain connection

If a time-slot is not used it is wasted

In ATM, each cell carries in it the identification of the virtual circuit it belongs to

This makes it possible to assign fixed capacity with a

constant delay to a virtual connection and have non-real-time traffic use the cells that were not used by the connection

ATM is first and foremost a technique for service differentiation:

Real-time voice and video can use Circuit Emulation

Services (CES) with a constant bit-rate (CBR) and delay

There are also real-time and non-real-time variable bit-rate (VBR) services for non-constant data streams

IP traffic can use unspecified bit-rate (UBR) services

All of these are implemented by switching standard ATM cells

(16)

ATM cell format

Flow Control – 4 bit

Virtual Path Identifier (VPI) – 8 bit

Virtual Channel Identifier (VCI) – 16 bit

Payload Type – 3 bit

Cell Loss Priority – 1 bit

Cyclic Redundancy Check (CRC) – 8 bit

0 1 2 3 4 5 6 7 Flow Control VPI (first 4) VPI (last 4) VCI (first 4)

VCI (last 4)

VCI (middle 8)

PL type P CRC

AAL header (0 or 4 bytes) payload (44 or 48 bytes)

bit

byte 1 2 3 4 5 6 - 53

(17)

ATM cell header

The Flow Control field was reserved for compatibility with IEEE 802.6 DQDB MAN and is always set to zero in the User- Network Interface (UNI)

In the Network-to-Network Interface (NNI) this field is used to carry 4 more bits of VPI the (the most significant 4 bits)

Consequently, VPI has 8 bits in UNI and 12 bits in NNI

The VCI field always has 16 bits

Payload type has three bits indicating the following:

user-generated cell

forward congestion notification

type indicator for higher layers (last cell in AAL 5)

Cell Loss Priority may be set by the ATM network to mark the packet ”discard-eligible” (DE)

The 8-bit CRC only covers the cell header

(18)

ATM addresses

An ATM address, as defined by the ATM Forum, is 20 octets long and has the following fields:

The Address Prefix, 13 octets, is interpreted hierarchically (left-to-right) and allows a variable number of hierarchy levels to be defined

The ID is 6 octets long and makes it possible to use IEEE 802 addresses

The Selector is a one-octet field used for subaddressing

There are three formats of 20-octet private ATM addresses:

NSAP Encoded E.164 format

DCC Format (Data Country Codes specified in ISO 3166)

ICD Format (International Code Designators identify particular international organizations)

PNNI routing is closely coupled with ATM addresses and they have to be assigned correctly (compare with CIDR)

(19)

ATM Service Classes

Constant Bit Rate (CBR) - specifies a fixed bit rate for Circuit Emulation (CE)

Variable Bit Rate - Real Time (VBR RT), provides a specified throughput with end-to-end synchronization but data is not sent evenly, used for real-time voice and video

Variable Bit Rate - Non-Real Time (VBR NRT), provides a specified throughput without end-to-end synchronization

Available Bit Rate (ABR) provides a guaranteed minimum capacity but allows data to be bursted at higher capacities when there is excess capacity in the network, ABR services are generally not implemented in ATM networks and services

Unspecified Bit Rate (UBR) does not guarantee any

throughput, used for non-time-critical applications, such as file transfers

(20)

ATM Adaptation Layers (AAL)

ATM can be used to provide very different types of services

To facilitate these services, the work of developing five

different ATM Adaptation Layers (AAL 1 through AAL 5) on top of ATM was started

Later AAL 3 and 4 were combined

The current ATM Adaptation Layers are:

AAL 1 - Constant Bit Rate (CBR) service (circuit emulation)

AAL 2 - Variable Bit Rate (VBR) service (voice and video)

AAL 3/4 - non-real-time VBR service

AAL 5 - for implementing Frame Relay service on ATM

AAL is divided into two sublayers:

Convergence sublayer

Segmentation and Reassembly (SAR) sublayer

(21)

ATM Standard Interfaces

ATMX

PNNI

ATMX ATM ATM X

host

(private) UNI

ATMX

PNNI

ATMX ATM X (public)

UNI

ATMX

ATMX ATM X

NNI Private ATM Network

Public ATM Network (operator 1) Public ATM Network

(operator 2)

(22)

User-Network Signaling

User-Network Signaling is a mechanism between the user and the network through which they:

Exchange control information

Request the use of network resources

Negotiate circuit parameters

The VPI/VCI pair as well as the requested bandwidth and

quality of service are allocated to a virtual circuit as a result of a successful signaling exchange

User-network signaling protocols support the opening and closing of connections

These signaling protocols are run on the Signaling ATM Adaptation Layer (SAAL)

(23)

ATM Signaling Protocol Stack

The figure below illustrates the protocol stack used in ATM signaling

Signaling ATM Adaptation Layer (SAAL) ensures the reliable delivery of signaling messages and consists of:

User-Network Interface Service Specific Coordination Function (UNI SSCF)

Service Specific Connection-Oriented Protocol (SSCOP)

ATM Adaptation Layer 5

User-Network Signaling UNI SSCF

SSCOP AAL 5

ATM

Physical Layer SAAL

(24)

ITU-T Q.2931

The ITU version of ATM signaling is an alternative to the ATMF version

The Q.2931 signaling protocol specifies the procedures for the establishment, maintenance and clearing of network connections at the Broadband ISDN (B-ISDN) UNI

The Q.2931 signaling procedures are defined in terms of messages exchanged (see message format below)

Message Header Information Element 1 Information Element 2

...

Information Element N

(25)

UNI 4.1

ATMF has defined a series of UNI specifications based on the ITU-T Q.2931 recommendation

The version currently implemented in most ATM devices is UNI 4.1

UNI 4.1 provides the signaling procedures for dynamically establishing, maintaining and clearing ATM connections at the ATM User-Network Interface

UNI 4.1 applies both to Public and Private UNI

UNI 4.1 is defined in the ATMF document: ATM User-Network Interface (UNI) Signaling Specification, version 4.1

UNI version 4.1 has a lot of added functionality (& complexity) compares with version 3.0 used in early ATM switches

(26)

PNNI Routing

The Private Network-to-Network Interface (PNNI) is a routing protocol with the following properties:

It is hierarchical

It is dynamic

It is an OSPF-like link-state protocol

PNNI was designed to scale to large ATM networks

PNNI has been tested in real use in the ATM network of Telecom Finland (now Sonera) since 1997

It has been found to work well in a network with several hundred nodes

It can recover from various error situations within a few seconds

(27)

Critique of ATM

The technology was developed for the next generation of the PSTN (namely B-ISDN) and is now being used for

internetworking which is something totally different

Even if the basic concept is very simple, the service contract and its negotiation at the UNI make ATM very complex

ATM specifications are too heavy for today’s needs:

There are too many AALs, AAL 1 and AAL 5 would be sufficient for most purposes

The ISDN-like ATM signaling is too complicated

The connection-oriented paradigm is too heavy for many purposes

Consequently, developing ATM equipment and applications is slow and expensive

With current programming methodology, it is next to

impossible to create programs that could adequately utilize the service options provided by ATM

Referências

Documentos relacionados

7.2.2.3 Actors for User Interaction within the Smart City Table 5: Example of actors End node-Users Connector/Gateways Network Servers Application Servers IoT Devices