3.1 Transport Layer Flashcards
Logical communication meaning
an end-to-end connection in which the sender and receiver are directly linked
Transport layer vs network layer
transport layer: logical communication between PROCESSES (apps)
- above the network layer
- uses port numbers
network layer: logical communication between HOSTS
- right below the transport layer
- uses IP addresses
TCP vs UDP services
TCP: Transmission Control Protocol
- reliable, in-order delivery
- congestion control
- connection setup
UDP: User Datagram Protocol
- unreliable, unordered delivery
TCP definition
connection-oriented, reliable full duplex protocol supporting a pair of byte streams (one in each direction)
“full duplex” - data is sent and received at the same time, like a 2-way street
Where is the transport layer implemented?
At the hosts (which are at the network edge)
Do UDP segments with the same port number always go to the same socket?
Yes, IP/UDP datagrams with the same destination port number will be directed to the same socket at the receiving host
- this is ALWAYS true, even if they have different source IP addresses and/or source port numbers
Do TCP segments with the same port number always go to the same socket?
Not necessarily– because there might be multiple different source IP addresses or source ports involved, the segments may go to different ports
- so, if TCP segments from different hosts are sent to the same port #, they may still go to different sockets
port number function
distinguishes different services/apps on a device
help direct network traffic to correct application/service
Port number definition
16-bit number used by the transport layer to identify specific apps running on a host
allows multiple apps/services on the same host to communicate at the same time
for example, port 80 is used for HTTP, and port 443 is used for HTTPS.
demultiplexing at the CLIENT vs SERVER
client: making sure the info is sent to the correct app on the host (since host is usually running many apps at once)
server: making sure HTTP messages are send to the correct client (since server is handling requests from many clients at once)
multiplexing at the CLIENT vs SERVER
client: funneling messages from apps into a packet
server: combining data from multiple sockets and adding transport header that’s later used for demultiplexing
demultiplexing
datagram’s payloads (the data inside the packet) are directed to the correct app/protocol in the host
apps/protocols are identified by a socket number
happens on all layers
multiplexing
the reverse of demultiplexing; as apps send messages down through different sockets, TCP funnels them down into an IP packet to be sent out
What does a port scanner do?
Checks for open ports on a server
What is one way that TCP packets can be verified for data integrity?
Compare the packet’s checksum
connectionless multiplexing/demultiplexing (UDP)
UDP sender takes each application-layer chunk of data written into a UDP socket and send it in a distinct UDP datagram
UDP datagram specifies:
- dest IP address
- dest port #
receiving host directs segment to the socket with that port #
UDP will deliver a segment’s payload into the appropriate socket, preserving the application-defined message boundary
What are 3 ways for a server to detect congestion?
- sender infers loss because of a missing ACK from the receiver
- server sends a packet indicating congestion
- sender measures RTTs and compares them to the current RTT measurement
connection-oriented demultiplexing (TCP) - the 4 parts of TCP sockets
more complicated than UDP
TCP sockets are identified by 4 things:
*source IP address
* source port number
*dest IP address
* dest port number
for demultiplexing, the receiver uses all 4 values to direct the segment to the right socket
the header of a TCP packet contains:
source port # and dest port #
no IP addresses!
What does the UDP header length field do
tells UDP where the segment ends, because segments vary in length
would a checksum detect a single flipped bit?
yes
would a checksum detect two flipped bits?
not necessarily–2 flipped bits might not change the checksum
What does a cumulative acknowledgement ACK do, as between sender and receiver?
Acknowledges all packets with a sequence # up to the ACK being received
What is selective repeat?
an error recovery method that resends only the packets that were dropped/received incorrectly
so, if the receiver is missing a packet with a sequence # lower than other received packets, the sender will realize this when it doesn’t receive an ACK for that packet
the sender will wait a certain amount of time, then resend the missing packet
Why does the ACK (sent by the receiver) add 1 to the sequence number of the received datagram?
the segment sent by the sender back to the receiver is only 1 byte
flow control definition
TCP mechanism to prevent congestion
where the receiver tells the sender how much free buffer space it has, it is NOT possible for the sender to send more data than the receiver has room to buffer
implemented on the sender side to prevent sending more data than the receiver can handle at once
TCP 3-way handshake codes
- SYN
- SYN-ACK
- ACK
