ECE361 Midterm 1: b. Computer Network Architectures

What are the 6 questions we should ask when defining a computer network architecture?

1. what services are provided by the network

2. what services are provided by entities attached to the network

3. how are resources shared to disseminate (spread) info

4. how are major hardware components of the system organized and connected

5. how is software organized

6. how can user processes running in remote computer communicate using a network infrastructure

what is a connection oriented network service?

one that establishes and path before sending and all data flows along the same path (FIFO)

what are the 3 aspects are needed for a connection oriented service

1. reliable data transfer: in order delivery, no duplicates or missing data

2. flow control: do not congest the receiver

3. congestion control: do not congest the networks

what is a connectionless network service

one that transmits packets independent of path and may follow different routes

what delivery and protocol method do connectionless services use?

1. no delivery guarantees

2. end-to-end protocol: each edge requests from the source and its only in charge of itself

   1. routers don’t need to handle flow control, only hosts

what is switching?

how info flow uses multiple nodes and links (routing data)

what is multiplexing?

how multiple info flows into the same link, sharing channel

what are switching methods used for?

allocation of shared network resources and the transport of user data

how does circuit switching work

1. circuit establishment: a dedicated circuit between intermediary switching centers

2. data transfer: data can be transferred between the source and destination once the circuit has been established

3. circuit disconnection: when one of the users initiates the disconnect

what are some of the limitations with circuit switching?

1. set up phase delay when establishing connections

2. signaling overhead when ending connections

= introduces a time delay and network traffic

what is multiplexing in circuit switching

sharing a given communication channel among multiple connections

what are the 2 types of multiplexing in circuit switching

1. frequency division multiplexing (FDM)

2. time division multiplexing (TDM)

how does frequency division multiplexing in the circuit switching work?

the frequency spectrum of a link (the medium) is partitioned into multiple bands. each connection get its own frequency band. some frequency bands have no connections

how does time division multiplexing in the circuit switching work?

time is divided into frames of fixed duration. frames are divided into fixed number of time slots. connections are assigned to a time slot. connections occupy the same time slot for multiple frames

what is the limitation of frequency division multiplexing (FDM) and time division multiplexing (TDM)

users can have silent periods of time, wasting resources

what is message switching

a message is sent as a single unit from one node to another without a dedicated path between the sender and receiver

how does message switching work

• message: has a header with information, used for forwarding a message to destination

• store and forward: each relay node must store the entire message before forwarding. transmission occurs only if the next link and node are available, otherwise, the message waits in storage until resources are free

• resources are shared among different calls (transmissions)

what is statistical multiplexing?

sharing the same communication channel/bandwidth among multiple connections without fixed allocations of the resource to those connections

what is the limitation to statistical multiplexing?

the entire message must be received at a switch before it can be forwarded

what is packet switching

1. message from the sender is broken into packets: each packet consists of a header and a payload

2. store and forward: a packet switch stores each packet it receives and determines how to forward it based on the header info

(resources are shared among connections)

what is the benefit of packet switching

packets that are from the same connection can be processed concurrently, saves time

what are datagrams

an approach to packet switching that avoid connection setup delay

how do datagrams work

1. packet has header with destination, no preestablished path

2. switch reads packet destination

3. switch checks routing table

   1. tells packet next hop based on destination

4. packet is forwarded, routed independently

multiple routes: can also be rerouted if there is an error

what is the limitation with datagrams

packets can be received at different sequences

what are virtual circuits

approach to packet switching with a preestablished path

how do virtual circuits work

1. the routing table at each switch is based on the preestablished path, like statistical multiplexing

2. the packet header specifies which virtual circuit to use

(relative or global names can be used, relative VC names need label switching)

what is ARPANET

first packet switching protocol, uses IMP (interface message processor), or the ARPANET router

what did the beginnings of protocol layering look like?

• end nodes didn’t know what was happening in the network (routing)

• only considered passive info transfer, not applications (active instructions)

• few destinations

how does the layering model work?

• partition services and functions into layers

• each layer of service is provided by peer protocol entities (nodes following the layers protocol)

• identifiers within protocols = identifiers across layers

what is the OSI architecture for protocol stacks?

breaks down flow of info

• application: end user services, user interaction

• presentation: formatting, encryption, compression of data

• session: setup and management of end-to-end dialogue and connection

• transport: end-to-end delivery of messages to processes, arrival in right order, reassmbles, retransmits

• network: end-to-end transmissions (of packets) in net, routing, assigns IP address

• link: makes sure data moves smoothly between devices of the same network (MAC and LLC)

• physical: transmission of bits over physical media

what is payload

data

how does encapsulation work at each layer

each layer needs this to process the payload from the previous layer

application: high level control information for request

transport: utilize datagrams, read control info

network: add source and destination for routing, go through routers, forward info from client → router → server

link: take network layer payload, deal with packets (IP addresses) → frames (MAC addresses)

physical: info bits are sent with a waveform receiver, decodes

what is the internetworking architecture?

application layer: http, email, etc.

transport services: routing, …

transport layer: UDP, TCP

network layer: IP, …

link layer: ethernet, wifi, optical fiber, …

what defines networking interconnection? (5)

• heterogeneous transmission media

• heterogeneous hardware & OS

• common address space used to identify where a host/router is connected to the internetwork

• common protocol for all gateways (hardware/software program that acts as an entry and exit point between different networks) and hosts

• common protocol used for data transfer and signalling

how was the internet a big ARPANET?

• destinations are attached to routers

• routers speak for destinations, one router per destination — assumes a destination has a single instance

• host to IMP → IP

• IMP to IMP → TCP

• host to host → UDP

router: “i want popcorn“

other router: “i am the only one whos has popcorn“

how does the ip internet work today?

• networks are organized into autonomous systems (different networks can run different protocols)

• routers run IP to forward data packets

• multiple routers for a source/destination network

router: “i want popcorn“

other router: “they have popcorn“ → points to network

ex. of internet use, how do you get a page from the web?

1. computer (client) and attached router are configured

2. host and router communicate through a LAN (local area network) or point to point link

3. data is broken into packets, router over links, computer for routing, and networks

4. host obtains IP address of remote server

5. client starts end to end reliable connection with remote server and starts exchanging messages

Must the Internet Protocol (IP) be used at the network of any computer network layer as packets flow across networks from their origins to their destination?

No, it is currently the norm as no translations are needed and it uses one network layer protocol

What advantage do you think is attained by using IP at the network layer of the networks that constitute the Internet?

Uses only IP for packet switching instead of multiple protocols

What is the importance of having a transport layer with respect to application developers

Application developers do not have to redesign the services provided by the transport layer for each new application

Why do you think routing tables today [for datagram-based packet switching] state simply the next hop towards an intended destination address?

memory was expensive