Communications in Computing. EE-3710

What are the types of delay

Transmission, Propagation, Processing, Queuing

Transmission delay

Time it takes to get packet data from host to transmission medium

Propagation delay

Time it takes for whole packet to reach destination

Transmission delay formula

packet length (L) / link bandwidth (R)

Propagation delay formula

length of physical link (d) / propagation speed (s)

nodal processing delay formula d_nodal

d_proc + d_queue + d_trans + d_prop

processing delay

time to check bit errors and determine output link

queuing delay

time waiting at output link for transmission, varies with congestion

queuing delay formula

packet length (L) * average packet arrival rate (a) / link bandwidth (R)

traceroute

measures delay by sending test packets and seeing how long it takes for them to come back

packet sniffing

reads packets as they’re transferred

DoS

denial of service from an overwhelming number of packets

TCP

establishes a connection prior to sending messages

advantages to TCP

reliable, flow control, congestion control

disadvantages to TCP

requires setup, slower, no minimum throughput guarantee

UDP

sends packets with IP address and port number attached, no prior connection established

advantages to UDP

faster,

disadvantages to UDP

unreliable, it can suffer packet loss, can’t control flow or congestion

HTTP

utilizes TCP for transport services, requests, receives, and displays web objects.

what standard port does HTTP use

port 80

what standard port does email use

port 25

what are “cookies”

it stores a key that’s used to access client-specific server information

what is round trip time (RTT)

time it takes to get from client to server and back.

what is dynamic adaptive streaming over HTTP (DASH)

divides video into multiple chunks, encodes at different rates

advantages of DASH

since the client can choose their own encoding rate it means delays and buffering can be avoided. Highest video quality still available so long as there’s bandwidth

disadvantages of DASH

since the video can be encoded at different rates those must be stored on the server, which costs space and computation power when switching encoding.

Enter Deep

large quantity of small servers

Bring Home

smaller number of large servers

advantages/disadvantages of Enter Deep

harder to maintain and less cost effective for ISP but it places servers closer to the client

advantages/disadvantages of Bring Home

easier to maintain and slightly cheaper but servers are farther from some clients

client push

primarily for clients to send data to a server (SMTP)

client pull

primarily for clients to receive data from the server (HTTP)

simple mail transfer protocol (SMTP)

client push, sends multiple objects at once, has status codes

what RFC defines syntax for email messages

RFC 2822

what is the domain name system (DNS)

structure of servers designed to resolve domain name disputes.

Root servers

last resort servers that handle especially difficult to translate names, managed by ICANN

Top Level Domains (TLD)

each responsible for a specific domain (.org, .net, .gov)

Authoritative servers

organizations’ own DNS server that translates domains within that organization

Local DNS server

where hosts go to resolve domain translation

iterated query

reaches out to every local DNS until it finds the authoritative DNS server

recursive query

burden of name resolution on the local servers, each contacting the next until the name gets resolved.

DDoS attack

coordinating flood of root servers, has never worked

CBR

constant bit rate

VBR

variable bit rate

multiplexing

adds transport header, handles data from multiple sockets

demultiplexing

use header info to deliver received segments to correct socket

connectionless demultiplexing

checks destination port number and directs it to appropriate socket (UDP)

connection demultiplexing

uses source, destination port numbers and IP address to direct to correct socket (TCP)

what is a finite state machine (FSM)

sends and receives data from underlying channel

what is the underlying channel

a perfect transport medium with no packet loss or bit errors

qualities of RTD 1.0

reliable underlying channel, FSM for both sender and receiver,

qualities of RDT 2.0

can flip bits in the event of bit errors, will resend packets that have errors

qualities of RDT 3.0

has a timer for ACKs, will re transmit if timer goes off. Utilizes sequence numbers to manage duplicates. has timeout feature for lost/corrupted packets

how does RDT 2.1 handle corrupted ACK/NAK

will resend packet with sequence number to avoid duplicate concerns

pipelining

sends multiple packets at once for better performance

qualities of go-back-N

window of un-ACKed packets, sequence numbers to keep track of packet order.

what happens in go-back-N when a packet fails

sender will resend every packet from the failed one onward, any successful packets after the failed one will be discarded and receiver will wait for new ones.

qualities of selective repeat

receiver individually ACKs packets as they’re received, timer for timeout/failure, window may store ACKed and un-ACKed packets.

what happens in selective repeat when a packet fails

failed packet gets resent, all successful transmissions after failure are held in buffer until failed packet is resent and ACKed

cumulative ACK

ACK number is the sequence number (SEQ) + data size

what is a 3-way handshake in TCP

data is sent, then receiver ACKs and the sender confirms that ACK with its own

additive increase

increase sending rate by 1 max segment size every RTT until loss

multiplicative decrease

cut sending rate in half every loss event

TCP tahoe

always sets congestion window (CWND) to 1 during timeout or duplicate situation

Goodput

rate of unique packets arriving

congestion window (CWND)

used to determine congestion, if window fills up then its congested