Bachelor_20062_Network :: 2006/10/28 00:29

/분류없음

Chapter 1 Computer Networks and the Internet

  1. What is Internet?
    1. Packet switching
      1. Allows multiple communicating end systems to share a path at the same time.
      2. routers & link-layer switches
    2. Internet Engineering Task Force(IETF)
      1. Internet standards are developed.
      2. Request For Comments(RFCs)
        1. standard document
    3. Protocol
      1. It defines the format and the order of messages exchanged between two or more communicating entities, as well as the actions taken on the transmission and/or receipt of a message or other event.
  2. The Network Edge
    1. Connection-oriented service
      1. Transmission Control Protocol(TCP)
      2. The client program and the server program send control packets to each other before sending packets with the actual data to be transferred.
      3. End systems are connected in a very loose manner
        1. Only end systems themselves are aware of this connection.
        2. The packet switches within the Internet are completely oblivious to the connection
      4. Reliable data transfer
      5. Flow control
      6. Congestion control
    2. Connectionless Service
      1. User Datagram Protocol(UDP)
      2. No handshaking
      3. Ideal for simple transaction-oriented applications
      4. No reliable data transfer
  3. The Network Core
    1. Circuit Switching
      1. The resources needed along a path to provide for communication between the end systems are reserved for the duration of the communication session.
      2. There is a bona fide connection for which the switches on the path between the sender & receiver
      3. Frequency-division multiplexing(FDM)
        1. The link dedicates a frequency band to each connection for the duration of the connection
      4. Time-division multiplexing(TDM)
        1. The network dedicates one time slot in every frame to the connection.
        2. It is wasteful because the dedicated circuits are idle during silent period.
    2. Packet Switching
      1. The resources are not reserved, a session's messages use the resources on demand.
      2. Store-and-Forward Transmission
        1. The switch must receive the entire packet before it can begin to transmit the first bit of the packet on to the outbound link.
      3. Output buffer
        1. Stores packets that the router is about to send into that link
      4. Queuing delay
        1. The arriving packet must wait in the output buffer if the link is busy.
      5. packet loss occur.
    3. Packet switching is better choice!
      1. It offers better sharing of bandwidth than circuit switching.
      2. It is simpler, more efficient, and less costly to implement.
      3. It provide higher speed when there are few users.
      4. Statistical multiplexing
        1. "On-demand sharing of resources"
    4. Packet-Switched Networks
      1. Virtual-Circuit Network
        1. Any network that forwards packet according to virtual-circuit number
        2. The actual address of the source & destination are not needed to perform switching
        3. Maintains state information
        4. 실제로는 circuit system 이 아닌데 packet system 을 가지고 circuit 을 흉내를 내보자. 즉 통로가 이미 정해져 있다는 것에서 circuit system 과 비슷한 면을 가진다.
      2. Datagram Network
        1. Each packet switch has a forwarding table that maps destination addresss to an outbound link
  4. Access Networks and Physical Media
    1. Access Network
      1. End system 을 edge router 로 연결한 link
    2. Edge router
      1. End system 이 다른 system 으로 가기 위한 path 에 있는 첫번째 router 이다.
    3. 기존 연결 문제점
      1. 속도 넘 느리다
      2. 전화선을 이용하기 때문에 인터넷을 단독으로 할 수 없다.
    4. 이리하여 등장!
      1. DSL
        1. 기존 전화선을 동일하게 사용
        2. Frequency 를 나누에 downstream, upstream, telephone 이렇게 사용가능
      2. HFC
        1. TV 동선 이용
    5. Physical Media
      1. Guided media
        1. waves are fuided along a solid medium
        2. fiber-optic, twisted-pair copper, coaxial cable
      2. Unguided media
        1. waves propagate in the atmosphere and in outer space
        2. wireless LAN, satellite channel
  5. ISPs and Internet Backbones
    1. Tier-1 ISP(Internet Backbone)
      1. Directly connected to each of the other Tier-1 ISPs
      2. Connected to a large number of Tier-2 ISPs & other customer network.
      3. International in coverage
    2. Tier-2 ISP
      1. Regional or national coverage
      2. Connects to only a few of the Tier-1 ISPs
  6. Delay and Loss in Packet-Switched Networks
    1. Processing delay
      1. The time required to
        1. examine the packet's header
        2. determine where to direct the packet
        3. check for bit error in the packet that occurred in transmitting the packet's bits from the upstream node to router
      2. After this nodal processing, the router directs the packet to the queue that precedes the link to router B
    2. Queuing delay
      1. Waiting time to be transmitted onto the link.
      2. 4 가지 delay 중 queuing delay 를 줄이는 게 가장 현실적이다. 나머지들은 hardware 와 관련이 깊지만 queuing delay 는 packet 의 순서를 조절함으로써 delay 를 어느정도 줄일 수 있다.
      3. router 에서 걸리는 queuing delay 시간은 평균값으로 판단. 각 packet 마다 기다리는 순서에 의해 queuing delay 가 다르기 때문에 평균값으로 이야기한다.
    3. Transmission delay(Store-and-Forward delay)
      1. Time required to push all of the packet's bits into the link
      2. Length of the packet / Transmission rate of the link
    4. Propagation delay
      1. The time required to propagate from the beginning of the link to router B
      2. It is equal to, or a little less than, the speed of light
      3. Distance between two router / Propagation speed
    5. Total nodal delay
      1. Process + Queuing + Transmission + Propagation
    6. End-to-End delay
      1. N * ( Process + Transmission + Propagation ),
      2. where N-1 routers between the source host and the destination host.
      3. Queuing delays are negligible.

Chapter 2 Application Layer

  1. Principles of Network Applications
    1. Network Application Architecture
      1. Client-Server Architecture
        1. There is an always-on host, called the server, which services requests from many other hosts, called clients.
        2. Clients do not directly communicate with each other
        3. Single server host is incapable of keeping up with all the requests from its clients -> clusters of host
      2. P2P Architecture
        1. There isn't an always-on server at the center of the application
        2. Arbitrary pairs of hosts,called peers, communicate directly with each other
        3. None of the participating hosts is required to be always on.
        4. It is intrinsically scalable.
      3. Hybrid Architecture
        1. Client-Server and P2P architecture
        2. Example) Instant message changing, like MSN.
    2. Process Communicating
      1. Socket : It is the interface between the application layer and the transport layer with in a host
      2. Port Number : For the purpose of the sending host must also identify the receiving process running in the host.
    3. Application service requirements along three dimension
      1. Reliable data transfer : Reliable data transfer VS loss tolerant app.
      2. Bandwidth : bandwidth-sensitive app VS elastic app.
      3. Timing : realtime app VS non-realtime app.
    4. Services Provided by the Internet Transport Protocols
      1. TCP
        1. Connection-oriented
          1. The Service is referred to as a "connection-oriented" service rather than an "connection" service because the two processes are connected in a very loose manner.
        2. Full duplex
        3. Congestion control
        4. Does not guarantee
          1. Minimum transmission rate
          2. Delay rate
      2. UDP
        1. Unreliable data transfer service
        2. Sending process can pump data into a UDP socket as any rate it pleases.
        3. Does not guarantee
          1. message will never reach the receiving process
          2. message may arriving out of order
          3. transmission rate & delay rate
  2. The Web and HTTP
    1. Stateless protocol
      1. The server send requested files to client without storing any state information about the client
    2. Nonpersistent & Persistent Connections
      1. HTTP use persistent connections with pipelining in its default mode
      2. Round Trip Time(RTT)
        1. The time it takes for a small packet to travel from client to server and then back to client.
        2. Propagation delay + Queuing delay + Processing delay.
      3. Nonpersistent
        1. Each TCP connection transports exactly one request message and one response message.
        2. Example
          1. Suppose the page consists of a base HTML file and 10 JPEG images, and that all 11 of these objects reside on the same server
          2. When a user requests the Web page, 11 TCP connections are generated
          3. How much time is needed?
            1. RTT + RTT + Transmission time at the server of the object
            2. First RTT
              1. The first two part of three way handshake take one RTT
            3. Second RTT
              1. HTTP request message combined with the third part of the three way handshaking + server send the HTML file into the TCP connection
      4. Persistent
        1. Shortcoming of Nonpersistent connection
          1. Brand-new connection must be established and maintained for each requested object
          2. Each object suffers a delivery delay of two RTTs
        2. The server leaves the TCP connection open after sending a response. Typically the HTTP server classes a connection when it isn't used for a contain time.
        3. Two versions of persistent
          1. Without pipelining
            1. The client issues a new request only when the previous response has been received.
          2. With pipelining
            1. HTTP client issues a request as soon as it encounters a references.
    3. HTTP Message Format
      1. Request Message
        1. Message is written is ordinary ASCII text.
        2. Each line followed by a carriage return & line feed.
        3. Method
          1. GET
            1. It is used when the browser requests an object, with the requested object identified in the URL field.
            2. Often used when inputting data in the requested URL.
          2. POST
            1. The entity body contains what the user entered into the form fields.
          3. HEAD
            1. It is similar to the GET method.
            2. It responds with an HTTP message but it leaves out the requested object
            3. Often used for debugging.
          4. PUT(HTTP 1.1 only)
            1. It allows a user to upload an object to a specific path on a Web server.
          5. DELETE(HTTP 1.1 only)
            1. To delete an object on a Web server
        4. Entity
          1. It is empty with the GET method, but is used with the POST method.
      2. Response Message
        1. Entity
          1. It contains the requested object itself.
    4. Cookies
      1. Background
        1. HTTP browser is stateless.
        2. It enables server high performance.
        3. But server doesn't know about client informations.
      2. Four component
        1. A cookie header line in the HTTP response message.
        2. A cookie header line in the HTTP request message.
        3. A cookie file kept on the user's end system and managed by user's browser.
        4. A back-end database at the Web site
      3. Scenario
        1. First visit to Web server using cookie
          1. Who the request comes into the Web server, the Web site creates a unique ID number and save ID in its database
          2. Server responds to client's browser, including in the HTTP response a "Set-Cookies:" header, which contains the id number.
          3. Client's browser receives the HTTP response message, it sees the "Set-Cookies:" header
          4. The browser then appends a line to the special cookie file that the browser manages. The line include the host name of the server & the ID number
          5. Each time client requests a Web page client's browser consults client's cookie file, extracts client's ID number for this site, and puts in the HTTP request a cookie header line that includes the identification number
        2. One week later, returns to that site
          1. Client browser will continue to put the header line "Cookie: ID" in the request message.
          2. Then server knows who comes into the server.
    5. Web caching
      1. How does it works?
        1. It also called a proxy server
        2. It is a network entity that satisfies HTTP request on the behalf of an origin Web server.
        3. The Web cache has its own disk storage and keep copies of recently requested objects in this storage.
        4. User's browsers can be configured so that all of the user's HTTP requests are first directed to the Web cache.
      2. The Conditional GET
        1. Mechanism that allows a cache to verify that its objects are up to date
        2. How is it requested?
          1. Request message uses the GET method.
          2. Request message include an "If-Modified-since" header line.
        3. Scenario
          1. Request for the first time.
            1. Proxy cache sends a request message to a Web server.
            2. Web server sends a response message with the requested object to the cache.
            3. The cache forwards the object to the requesting browser but also caches the object locally.
          2. One weeks later.
            1. Client request the same object via the cache, and the object is still in the cache.
            2. The proxy performs an up to date check by issuing a conditional GET.
            3. If the object has been modified since then, server send that object to the proxy. If not, server does not send that object and proxy forward its cached copy of the object to the client.
  3. File Transfer : FTP
    1. Common characteristic with HTTP
      1. They both run on top of TCP
    2. Differences
      1. FTP uses two parellel TCP connections to transfer file
        1. control connection
          1. used to send a control information, port 21
        2. data connection
          1. used to send a file, port 20, non-persistent
      2. Control connection open throughout the duration of the user session, but new data connection is created for each file transferred within a session.
    3. Out-of-band
      1. FTP uses a separate control connection
    4. In-band
      1. HTTP uses a same connection to transfer information & data
  4. Electronic Mail in the Internet
    1. SMTP
      1. It transfers message from senders' mail servers to the recipients' mail servers
      2. It restricts the body of all mail message to simple 7 bit ASCII.
      3. It does not normally use intermediate mail servers for sending mail, ever when two mail servers are located at opposite ends of the world.
      4. Persistent-connection : If sending mail server has several messages to send to the same receiving mail server, it can send all of the messages over the same TCP connection.
    2. Scenario
      1. Client SMTP has TCP establish a connection on port 25 to the server SMTP
      2. Server and client perform same application-layer handshaking
      3. During this handshaking, the SMTP client indirect the email address of the sender and email address of the recipient.
      4. Client send message. If it has messages to send to the server repeat send message.
      5. Close TCP connection.
    3. Comparison with HTTP
      1. Common
        1. Transfer file on TCP
      2. Difference
        1. Push & pull protocol
          1. Push protocol- SMTP(요청하는 쪽이 자료를 push 한다.)
          2. Pull protocol - HTTP(요청하는 쪽이 자료를 pull 한다.)
    4. Mail Message Format
      1. Multipurpose Internet Mail Extensions(MIME)
        1. To send content other then ASCII text, the sending user agent must include additional headers in the message.
        2. "Content-Type:" header
          1. Allows the receiving user agent to take an appropriate actions on the message.
        3. "Content-Transfer-Encoding" header
          1. Alerts the receiving user agent that the message body has been ASCII-encoded and indicates the type of encoding used.
      2. The received message
        1. The receiving server appends a "Received:" header line to the top of the message.
          1. From : The name of the SMTP server that sent message.
          2. By : the name of the SMTP server that received message.
          3. When : The time at which the receiving server received message.
    5. Mail Access Protocol
      1. Why do we need on the protocol?
        1. We can't use SMTP to obtain the messages because obtaining the message is a pull operation, whereas SMTP is a push protocol
        2. We need a special mail access protocol
      2. Post Office Protocol-Version3(POP3)
        1. Port 110, TCP connection
        2. Three phase
          1. Authorization
          2. Transaction
          3. Update
        3. "Download and delete"
        4. "Download and keep"
      3. Internet Mail Access Protocol(IMAP)
        1. It is a mail access protocol, but it has many more features than POP3
        2. Features
          1. Allow users to create folders and more message from one folder to another.
          2. Allow users to search message
          3. Maintain user state information
          4. Permit a user agent to obtain component of message
  5. DNS : The Internet's Directory Service
    1. DNS
      1. Translate hostname to IP address
      2. A distributed database in a hierarchy of DNS servers.
      3. A application-layer protocol that allows hosts to query the distributed database
      4. Run over UDP & uses port 53
      5. Few other important services
        1. Host aliasing
          1. DNS can be invoked by an application to obtain the canonical hostname for a supplied alias hostname as well as IP address of the host
        2. Mail server aliasing
          1. DNS can be invoked by an mail application to obtain the canonical hostname for a supplied alias hostname as well as IP address of the host
        3. Load distribution
          1. DNS is used to perform load distribution among replicated servers such as replicated Web servers
    2. Problems with a centralized design
      1. A single point of failure.
      2. Traffic volume
      3. Distant centralized database
      4. maintenance
    3. A Distributed, Hierarchical Database
      1. Root DNS servers
        1. In the Internet there are 13 root DNS servers
        2. Each "server" is actually a cluster of replicated servers
      2. Top-level Domain(TLD) server
        1. These are responsible for top-level domain such as com, org, net, gov ..
      3. Authoritative DNS server
        1. Every organization with publicly accessible hosts on the Internet must provide publicly accessible DNS
      4. Local DNS server(also called, default name server)
        1. When a host makes a DNS query the query is sent to the local DNS server, which acts a proxy, forwarding the query into the DNS server hierarchy.
    4. Scenario
      1. Requesting host -> Local DNS server
        (asking www.cnn.com)
      2. Local DNS server -> Root DNS server,
        Root DNS server -> Local DNS server
        (return IP of TLD server, .com)
      3. Local DNS server -> TLD DNS server,
        TLD DNS server -> Local DNS server
        (return IP of Authoritative DNS server, dns.cnn.com)
      4. Local DNS server -> Authoritative DNS server,
        Authoritative DNS server -> Local DNS server
        (return IP of dest server, www.cnn.com)
      5. * The query from the requesting host to the local DNS server is recursive, and the remaining queries are iterative.
    5. DNS caching
      1. It can cache the information in the reply in its local memory
      2. Hosts and mappings between host names and IP addresses are by no means permanent. DNS servers discard cached information after a period of time(often set to two days)
    6. DNS record
      1. Resource Records(RR)
        1. Provide hostname to IP address mapping.
        2. Each DNS reply message carries one or more resource records.
      2. (Name, Value, Type, TTL)
        1. TTL is time to live of the RR
        2. The meaning of Name and Value depend on type
          1. Type=A
            1. Provides standard hostname to IP address mapping
            2. Name : host name
            3. Value : IP address
          2. Type=NS
            1. Name : domain
            2. Value : host name of an authoritative DNS server
          3. Type=CNAME
            1. This record can provide querying hosts the canonical name for a hostname
            2. Name : domain
            3. Value : canonical hostname for the alias hostname
          4. Type=MX
            1. MX records allow the host names of mail servers to have simple aliases
            2. Name : domain
            3. Value : canonical name of a mail server that has an alias hostname
      3. If DNS server is authoritative
        1. then the DNS server will contain a 'Type A' record for the hostname
      4. If DNS server is not authoritative
        1. then the server will contain a 'Type NS' record for the domain
        2. It will also certain 'Type A' record that produces the IP address of the DNS server in the Value field of the NS record
        3. Ex) The edu TLD server
        4. (umass.edu, dns.umass.edu, NS)
        5. (dns.umass.edu, 129,119,40,111, A)
    7. DNS message
      1. Both query and replay message have the same format
      2. Refer to p.133 on Textbook
    8. Inserting records into the DNS database
      1. Registrar
        1. It is commercial entity that verifies the uniqueness of the domain name, enters the domain name into the DNS database
      2. When you register the domain name 'networkutopia.com'
        1. you need to provide the name, IP of your authoritative DNS servers.
        2. Then registrar would then make sure that a 'Type NS' and a 'Type A' record are entered into the TLD com server.
        3. (networkutopia.com, dns1.networkutopia.com, NS)
        4. (dns1.networkutopia.com, 212.212.212.1, A)
    9. P2P File Sharing
      1. Centralized Directory
        1. Napster
        2. Use a large sever to provide the directory service
        3. When an active peer obtain a new object, or removes an object, it informs the directory server
        4. Send message periodically to the peers to see if they respond
      2. Query Flooding
        1. Gnutella
        2. Does not use a centralized server
        3. Querying flooding
          1. Forward the query message to all of their neighbors which in turn forward the message to all of their neighbors.
          2. Limited scope query flooding
      3. Exploiting heterogeneity
        1. KaZaA
        2. Napster + Gnutella
        3. Does not use a dedicated server
        4. Not all peers are equal. The more powerful peers are designated as group leaders and have greater responsibilities
        5. Features
          1. Requesting queuing
          2. Limit the number of simultaneous uploads to any value
          3. Incentive priorities
          4. Parallel downloading

Chapter 3 Transport Layer

  1. Introduction and Transport-Layer Services
    1. Transport Layer
      1. Provides for logical communication between application processes running on different hosts.
    2. Logical communication
      1. From an application's perspective it is as if the hosts running the processes were directly connected; in reality, the host may be an opposite side, of the planet, connected via numerous routers and a side rounge of link types.
      2. Network routers act only on the network-layer fields of the datagram.
    3. Relation ship between Transport and Network Layers
      1. Transport Layer : Logical communication between processes
      2. Network Layer : Logical communication between hosts
    4. TCP & UDP
      1. TCP
        1. Reliable data transfer.
        2. Congestion control
        3. Flow control
        4. ACK, SEQ, TIMER
      2. UDP
        1. Unreliable data transfer.
      3. Services are not available.
        1. Delay guarantee.
        2. Bandwidth guarantee.
    5. Terminology(in this Textbook)
      1. Segment : TCP, UDP packet
      2. Datagram : Network-layer packet
    6. About Network layer
      1. Best effort service, but it makes no guarantees.
      2. Does not guarantee
        1. segment delivery
        2. orderly delivery of segments
        3. integrity of the data in the segment
  2. Multiplexing and Demultiplexing
    1. Multiplexing and Demultiplexing
      1. Multiplexing
      2. The job of gathering data chunks at the source host from different sockets, encapsulating each data chunk with header information(that will later be used in demultiplexing) to create segments, and passing the segments to the network layer.
      3. Demultiplexing
      4. Delivering the data in a transport-layer segments to the correct socket.
    2. How does it works?
      1. Host uses IP address & port numbers to direct segment to appropriate socket.
      2. Two connections
        1. Connection less Multiplexing and Demultiplexing
          1. UDP socket is fully identified by a two tuple, source port + dest port
          2. If two UDP segments have different source IP and/or source port numbers, but have the same destination IP address and destination port number, then two segments will be directed to the same destinations process via the same destination socket.
        2. Connection-Oriented Multiplexing and Demultiplexing
          1. TCP socket is identified by a four tuple, source IP & port + dest IP & port
          2. If same port number and different source IP, these are demultiplexed on different sockets.
  3. Connectionless Transport : UDP
    1. Benefits of UDP
      1. Finer application-level control over what data is sent and where.
      2. No connection establishment
      3. No connection state
      4. Small packet header overhead
    2. UDP Segment Structure
      1. The UDP header has only four fields, each consisting of two byte.
      2. source port, dest port, length, checksum
    3. The UDP checksum
      1. Provides for error detection
      2. Why checksum is needed in UDP(unreliable protocol)?
        1. To insure header informations. Changes of dest port could make serious security problem.
      3. Why UDP provides a checksum in the first place, as many link-layer protocols also provide error checking?
        1. No guarantee that all the links between source and destination provide error checking. That is one of the links may use a link-layer protocol that does not provide error checking.
        2. Possible that bit errors could be introduced when a segment is stored in a router's memory.
  4. Principles of Reliable Data Transfer
    1. Reliable Data Transfer
      1. Stop and wait protocol
        1. Sender sends one packet, then waits for receiver response.
      2. rtd1.0 : Reliable data transfer over perfectly reliable channel
      3. rtd2.0 : Reliable data transfer over Channel with Bit Errors
        1. Three additional protocol capabilities are required.
          1. Error detection
          2. Receiver feedback
          3. Retransmission
        2. Two problems
          1. What happens if ACK/NAK corrupted?
          2. Duplicate packet. It cannot know whether an arriving packet contains new data or is a retransmission!
      4. rtd2.1 : Reliable data transfer over perfectly reliable channel
        1. SEQ number added to packet -> handle duplicate packet
      5. rtd2.2 : Reliable data transfer over perfectly reliable channel
        1. NAK-free solution
      6. rtd3.0 : Reliable data transfer over Lossy Channel with Bit Errors
        1. Sender waits "reasonable" amount of time for ACK and then retransmission!!
    2. Performance of rdt3.0
      1. It is a stop and wait protocol
      2. Utilization : The fraction of time the sender is actually busy sending bits into the channel
      3. Example
        1. 1Gbps(10^9 bit), 30 millisecond RTT, 1000 byte packet
        2. Transmission delay = (packet length in bits) / (transmission rate) = 8000 / 10^9 = 8 microsecond
        3. Utilization of the sender = 0.008 / 30.0008 = 0.00027
    3. Pipelined protocol
      1. Rather than operate in a stop-and-wait manner, the sender is allowed to send multiple packets without waiting for acknowledgements
      2. Two basic approaches toward pipelined error recovery can be identified.
      3. Go-Back-N & Selective Repeat
    4. Go-Back-N
      1. The sender is allowed to transmit multiple packets without waiting for an acknowledgment, but is constrained to have no more than some maximum allowable number,N, of unacknowledged packets in the pipeline.
      2. Variables
        1. N : Referred to as the windows size and the GBN protocol itself as a Sliding-window protocol.
        2. base : The sequence number of the oldest unacked packet
        3. nextseqnum : The sequence number of the next packet to be sent
        4. [0, base-1] : packets that have already been transmitted and acked.
        5. [base, nextseqnum - 1] : packets that have been transmitted but not acked.
        6. [nextseqnum, base + N - 1] : packets that can be sent immediately
      3. Windows size
        1. [0, 2^k - 1], if k is the number of bits in the packet sequence number field.
      4. The receiver discards out of packet
        1. Pros : simplicity of receiver buffering. receiver need not buffer any out of order packets.
        2. Cons : more retransmission would be required.
      5. Cumulative acknowledgment
        1. Indicating that all packets with a sequence number up to and including n have been correctly received at the receiver.
      6. GBN 에서 중요한 두가지 point
        1. Cumulative ack
          1. 즉 n 까지 확실하게 받았으니 다음것은 n+1 이겠거니 하고 기다린다는 말이다. 실제 TCP 는 이렇게 동작하지 않고 'n+1 을 기다리고 있다' 라고 ack 를 준다.
        2. 처음의 seq num 은 1 이다.
          1. 이게 0 부터 시작이 아니고 1 인 이유는 0 부터 시작시 cover 할 수 없는 상황이 생긴다. 예를 들어 sender 에서 seq 0 을 보냈는데 receiver 에서 잘못 받게 되어 ack 0 을 보내게 되면 cumulative ack 이기 때문에 server 에서는 0 번까지 제대로 받을 줄 안다. 잘못 보낸 0 번 packet 에 대해 회복할 방법이 없다. sever 의 seq 를 1 부터 시작하고 receiver 의 ack 는 0 부터 시작하면 이 문제는 해결된다.
      7. Three types of events
        1. Invocation from above
        2. Receipt of an ACK
        3. A timeout event
    5. Selective Repeat
      1. Problem of GBN
        1. A single packet error can cause GBN to retransmit a large number of packets, many unnecessarily.
      2. Selective Repeat
        1. avoid unnecessary retransmission by having the sender retransmit only those packets that it suspects were received error at the receiver.
      3. Actions taken under SR.
        1. The SR receiver will acknowledge a correctly received packet whether or not it is in order.
        2. The SR receiver reackonwledge received packets with certain sequence number below the current windows base.
      4. Window size
        1. N = 2^(k-1)
  5. Connection-Oriented Transport : TCP
    1. TCP
      1. connection oriented
        1. before proceeding transfer, the two process must first "handshake" with each other.
        2. TCP "connections" is not end-to-end TDM of FDM circuit. connection state resides entirely in the two end systems
      2. Full duplex service
        1. data flows concurrently at each sides.
      3. Point-to-Point
        1. A single sender and a single receiver
      4. MSS
        1. The maximum amount of application layer data in the segment, not the maximum size of the TCP segment including headers.
        2. Example) When TCP sends a large file, such as an image as part of a web page, it typically breaks the file in to chunks of size MSS(except for the last chunk, which will often be less than the MSS).
      5. MTU
        1. The length of the largest link-layer frame
    2. TCP segment structure
      1. Refer to p.211 on Textbook
      2. SEQ & ACK number
        1. SEQ : byte stream number of the "first byte in the segment".
        2. ACK : sequence number of the next byte Host A is expecting from Host B.
      3. Cumulative ACK
        1. TCP only acknowledge bytes up to the first missing byte in the stream.
      4. Selection of Sequence Number
        1. TCP connection randomly choose an initial SEQ number
        2. To minimize the possibility that
          1. a segment that is still present in the network from an earlier
          2. already terminated connection between two hosts is mistaken for a valid segment in a later connection between those same hosts.(which also happened to be using the same port numbers as old connection)
      5. Piggybacked acknowledgement
        1. ACK for client-to-server data is carried in a segment carrying server-to-client data
        2. data 를 보내면서 ACK 를 data 위에 얹혀 보내는 것.
  6. Principles of Congestion Control
  8. TCP Congestion Control

Trackback(0) : Comment(0)
Trackback Address :: http://kbckbc.mireene.co.kr/tatter/kbckbc/trackback/58
Name
Password
Homepage
Secret
< PREV |  #1 ... #808 #809 #810 #811 #812 #813 #814 #815 #816 ... #838  | NEXT >