--- # System prepended metadata title: CSE350期中筆記 tags: [NSYSU] --- # CSE350期中筆記 ###### tags: `NSYSU` ## Chapter1 1. Packet switching vs Circuit switching 2. Internet Delay & packet lost 3. Protocal stack & OSI model 4. Encapsulation ## Chapter2 1. Client & Server + 伺服器端有固定的 IP 位址且永遠開啟並等候要求，通常單一伺服器無法容納所有用戶端的請求，所以一個資料中心 (Data Center) 有許多主機來建立一個虛擬伺服器。 2. P2P(Peer to Peer) architecture: + 利用一對對間歇性連線主機 (Peer) 之間進行的直接通訊。 + Peer 不屬於服務提供者，所以不會透過專門的伺服器進行通訊。 + Self-scalability: Peer 之間會透過散佈檔案傳給其他 Peer 而增加系統的服務容量。 + Advantage: 通常不需要大量的伺服器基礎設施和伺服器頻寬。 + Challenge: + 對 ISP 友善：P2P 會將視訊串流與檔案傳輸應用等上傳資料流從伺服器移向家用 ISP而帶給 ISP 壓力。 + 安全性:因高度分散與開放的本質。 + 誘因:仰賴使用者貢獻。 + 由於每個peer不一定會在線上，所以會有管理上的困難 3. Transport service requirements: Common apps(舉例) + data integrity + security + timing + throughput 4. TCP VS UDP (PPT 2-15) + TCP + Properties: + reliable data transfer + flow control + congestion control + connection-oriented + Defact: Don't provide + timing guarantee + minimum throughput guarantee + security + UDP + Property: + unreliable data transfer + Defact: Don't provide + reliability + flow control + congestion control + timing, throughput guarantee + security + connection setup 5. HTTP is "stateless" + server maintains no information about past client requests + state壞處:程式會變得很複雜 6. HTTP non-persistent figure & persistent + non-persistent HTTP + at most one object sent over TCP connection + connection then closed + downloading multiple objects required multiple connections + Round-Trip Time(RTT) def: time for a small packet to travel from client to server and back. + Response time(downloading an object) + one RTT to initiate TCP connection + one RTT for HTTP request and first few bytes of HTTP response to return + file transmission time + Total: 2RTT + file transmission time + Defact: + 2 RTT/object + OS overhead for each TCP connection + persistent HTTP + multiple objects can be sent over single TCP connection between client, server. + one RTT for all the referenced objects. 7. Proxy sever: soft state + Proxy server(web cache)會保存最近被請求過的物件副本， 如果使用者請求的物件沒有在proxy server裡的話，proxy server會向原本的server請求物件 + soft state: 保存最近被請求過的物件副本，過一段時間就要 重新跟原本server請求物件 8. Pull & Push + Pull: Client主動向Server發出request(如:HTTP) + Push: Server主動向Client傳輸資料(如:SMTP) 9. DNS records + RR format:(name, value, type, ttl) + type A + name = hostname + value = IP address + type NS + name = domain + value = hostname of authoritative name server for this domain. + type CNAME + name = alias name for some canonical(the real) name + value = canonical name + type MX + value = name of mailserver associated with name ## Chapter3 1. Multiplexing & Demultiplexing(PPT 3-11) + Multiplexing at sender + Handle data from multiple sockets, add transport header. + Demultiplexing at receiver + Use header info to deliver received segments to correct socket. + How works + Host receives IP datagrams and uses IP addresses & port numbers to direct segment to appropriate socket. ![](https://i.imgur.com/EwRvfCB.png) 2. Checksum: 1's complement + Treat segment contents, including header fields, as sequence of 16-bit integers. ![](https://i.imgur.com/f5xvSN9.png) 3. Reliable Data Transfer(RDT) ![](https://i.imgur.com/oiaQVXU.png) + RDT1.0: reliable transfer over a reliable channel + underlying channel perfectly reliable + no bit errors + no loss of packets ![](https://i.imgur.com/CVK33iT.png) + RDT2.0: channel with bit errors + underlying channel may flip bits in packet, but won't loss packets + checksum to detect bit errors + Recover from errors + ACKs + NAKs + New features + error detection + receiver feedback ![](https://i.imgur.com/am8G6Ph.png) + RDT2.1: + sender + sender retransmits current pkt if ACK/NAK corrupted + sender adds sequence number to each pkt ![](https://i.imgur.com/f1u7wOM.png) + receiver + receiver discards (doesn’t deliver up) duplicate pkt ![](https://i.imgur.com/EFp17nG.png) + RDT2.2: A NAK-free protocol + using ACKs only + receiver sends ACK for last pkt received OK + receiver must explicitly include seq # of pkt being ACKed + duplicate ACK at sender results in same action as NAK: retransmit current pkt + RDT3.0: + Reason + underlying channel can also lose packets. + Approach + sender waits reasonable amount of time of ACK + retransmits if no ACK received in this time + if packet (or ACK) just delayed: + retransmission will be duplicate, but seq #'s already handles this. + receiver must specify seq# of pkt being ACKed. + requires countdown timer + Sender ![](https://i.imgur.com/Oef9Oex.png) 4. (PPT 3-45) 為何$U_{sender} = \frac{3L/R}{RTT + L/R}$中$RTT$不用乘3 ![](https://i.imgur.com/AgwtR8Q.png) ANS: $lim_{n\to\infty} \frac{n(3L/R)}{n(RTT+L/R)+2L/R} = \frac{3L/R}{RTT+L/R}$ 5. Go-Back-N + sender + ACK(n):ACKs all pkts up to, including seq#n -> cumulative ACK + may receive duplicate ACKs(see receiver) + timer for oldest in-flight pkt + timeout(n): retransmit packet n and all higher seq# pkts in window ![](https://i.imgur.com/iRTuOwJ.png) + receiver + always send ACK for correctly-received pkt with highest in-order seq# + may generate duplicate ACKs + need only remember expectedseqnum + out-of-order pkt + discard(don't buffer) + re-ACK pkt with highest in-order seq# ![](https://i.imgur.com/Scu14BD.png) 6. (PPT 3-52 下圖有錯) ![](https://i.imgur.com/9VarLQZ.png) 圖中紅框部分: 上圖尚未送出封包下圖卻已經ACK 7. selective repeat sender & receiver + sender + case1: 收到上層的資料 + 當收到上層的資料，sender檢查下一個可以傳的封包序號， 如果序號在窗格裡則把資料封裝成封包並送出。 + 如果序號沒有在窗格裡則存入緩衝區或者退回上層，稍後再傳。 + case2: Timeout + 每個傳出去的封包都會有一個計時器 + 如果該封包的計時器超過一定時間，則重傳該封包並重新開始 計時器 + case3: 收到ACK(n) + 收到序號n的ACK則把序號n的封包標記為已收到 + 如果序號n在窗格內且序號等於send_base則將 send_base設為序號最小的未確認封包上。 + receiver + case1: 接收到的封包序號位於\[rcv_base, rcv_base+N-1\] + 回傳該封包的ACK給sender + 若該封包之前沒收到過則存入緩衝區 + 若該封包的序號等於rcv_base，則之前存入緩衝區的封包 都往上層傳(與rcv_base連續的封包) + case2: 接收到的封包序號位於\[rcv_base-N, rcv_base-1\] + 回傳該封包的ACK給sender且不存入緩衝區 + Default: 不做任何事 8. TCP sender & receiver + sender + events + data rcvd from app + create segment with seq# + seq# is byte-stream number of first data byte in segment + start timer if not already running + timeout + retransmit segment that caused timeout + restart timer + ack rcvd + if ack acknowledges previously unacked segments + update what is known to be ACKed + start timer if there are still unacked segments 9. TCP Ack generation | event | TCP receiver action| | -------- | -------- | | 按預期seq編號到達，所有預期編號之前的區段皆已ACK | 等待下一區段到達再ACK，至多等待500ms | | 按預期seq編號到達，有區段尚未ACK | 立刻送出一筆ACK，一次確認兩筆區段 | | 比預期seq編號大的區段到達，出現Gap | 立刻送出duplicate ACK，指出預期seq編號 | | 抵達區段可填補Gap | 立刻送出ACK，指出Gap底部 | 10. TCP fast retransmit:網路不算差時可以快速重傳 + 逾時重送的機制，等待timeout會拖延重送遺失封包 + 利用"Duplicate ACK"來偵測封包的遺失 + sender通常一次送出大量區段，當區段遺失時便會收到許多相同的duplicate ACK + 當sender收到3個以上相同data的ACK，就會認為區段已遺失，不需等待timeout即重送 11. TCP 3-way handshake ![](https://i.imgur.com/Qg6450R.png) 12. TCP slow-start 變化圖 ![](https://i.imgur.com/rAq6qdp.png) 13. TCP Congestion control FSM diagram ![](https://i.imgur.com/rBGbGGv.png) 14. TCP 不公平的原因 + 假設只有TCP連線會通過瓶頸連結且所有連線都有相同的RTT + 每對主機與目的端之間都只存在單筆TCP連線 + RTT較短會有更高的產出率 15. TCP close connection ![](https://i.imgur.com/ze2wtYp.png)