Transcript l5-1 - Heyook Lab

Transport Protocols
Relates to Lab 5. UDP and TCP
1
Midterm
2
Roadmap
• UDP
– Unreliable, connectionless datagram service
• TCP
– Reliable, in order, connection-oriented, byte stream service
• Principles
– Multiplexing/demultiplexing
– How to build reliable service on top of unreliable service
3
Orientation
• We move one layer up and look at the transport layer.
User
Process
User
Process
User
Process
TCP
User
Process
Application
Layer
UDP
Transport
Layer
ICMP
IP
IGMP
Network
Layer
ARP
Hardware
Interface
RARP
Link Layer
Media
4
Orientation
• Transport layer protocols are end-to-end protocols
• They are only implemented at the hosts
HOST
HOST
Application
Application
Transport
Transport
Network
Data Link
Network
Data Link
Network
Data Link
Data Link
5
Transport Protocols in the Internet
• The most commonly used transport protocols are UDP and
TCP.
•
•
•
•
•
•
UDP - User Datagram Protocol
datagram oriented
unreliable, connectionless
simple
unicast and multicast
useful only for few applications,
e.g., multimedia applications
used a lot for services
– network management
(SNMP), routing (RIP),
naming (DNS), etc.
•
•
•
•
•
TCP - Transmission Control
Protocol
byte stream oriented
reliable, connection-oriented
complex
only unicast
used for most Internet
applications:
– web (http), email (smtp), file
transfer (ftp), terminal (telnet),
etc.
6
UDP - User Datagram Protocol
• UDP supports unreliable transmissions of datagrams
– Each output operation by a process produces exactly one
UDP datagram
• The only thing that UDP adds is multiplexing and
demultiplexing
• Protocol number: 17
Applications
Applications
UDP
UDP
IP
IP
IP
IP
IP
7
UDP Format
IP header UDP header
20 bytes
UDP data
8 bytes
Source Port Number
Destination Port Number
UDP message length
Checksum
DATA
0
15 16
31
• Port numbers identify sending and receiving applications (processes).
Maximum port number is 216-1= 65,535
•Message Length is at least 8 bytes (I.e., Data field can be empty) and at
most 65,535
•Checksum includes UDP header and data.
8
Port Numbers
• UDP (and TCP) use port numbers to identify applications
• A globally unique address at the transport layer (for both UDP
and TCP) is a tuple <IP address, port number>
• There are 65,535 UDP ports per host.
User
Process
User
Process
User
Process
User
Process
TCP
User
Process
UDP
IP
User
Process
Demultiplex
based on
port number
Demultiplex
based on
Protocol field in IP
header
9
Transport Control Protocol (TCP)
10
Overview
Byte Stream
Byte Stream
TCP = Transmission Control Protocol
• Connection-oriented protocol
• Provides a reliable unicast end-to-end byte stream over an
unreliable internetwork.
TCP
TCP
IP Internetwork
11
Connection-Oriented
• Before any data transfer, TCP establishes a connection:
• Analogy: making a phone call
• One TCP entity is waiting for a connection (“server”)
• The other TCP entity (“client”) contacts the server
• Each connection is full duplex
CLIENT
SERVER
Request a co
nnection
onnection
Accept a c
Data Transer
waiting for
connection
request
Disconnect
12
Reliable
• Byte stream is broken up into chunks which are called segments
• Receiver sends acknowledgements (ACKs) for segments
• TCP maintains a timer. If an ACK is not received in time,
the segment is retransmitted
•Detecting errors and packet losses:
• TCP has checksums for header and data. Segments with
invalid checksums are discarded
• Each byte that is transmitted has a sequence number
13
Byte Stream Service
• To the lower layers, TCP handles data in blocks, the
segments.
• To the higher layers TCP handles data as a sequence of
bytes and does not identify boundaries between bytes
• So: Higher layers do not know about the beginning and
end of segments !
Application
Application
1. read 40 bytes
2. read 40 bytes
3. read 40 bytes
1. write 100 bytes
2. write 20 bytes
TCP
queue of
bytes to be
transmitted
Segments
TCP
queue of
bytes that
have been
received
14
TCP Format
• TCP segments have a 20 byte header with >= 0 bytes of data.
IP header TCP header
20 bytes
TCP data
20 bytes
0
15 16
Source Port Number
31
Destination Port Number
Sequence number (32 bits)
header
length
0
Flags
TCP checksum
20 bytes
Acknowledgement number (32 bits)
window size
urgent pointer
Options (if any)
DATA
15
TCP header fields
• Port Number:
• A port number identifies the endpoint of a connection.
• A pair <IP address, port number> identifies one
endpoint of a connection.
• Two pairs <client IP address, server port number>
and <server IP address, server port number> identify
a TCP connection.
Applications
Ports:
23 80 104
Applications
7
80 16
TCP
TCP
IP
IP
Ports:
16
TCP header fields
• Sequence Number (SeqNo):
– Sequence number is 32 bits long.
– So the range of SeqNo is
0 <= SeqNo <= 232 -1  4.3 Gbyte
– The sequence number in a segment identifies the first
byte in the segment
– Initial Sequence Number (ISN) of a connection is set
during connection establishment
17
TCP header fields
• Acknowledgement Number (AckNo):
– Acknowledgements are piggybacked, I.e
a segment from A -> B can contain an
acknowledgement for a data sent in the B -> A direction
– A hosts uses the AckNo field to send acknowledgements.
(If a host sends an AckNo in a segment it sets the “ACK flag”)
– The AckNo contains the next SeqNo that a host is
expecting
Example: The acknowledgement for a segment with
sequence numbers 0-1460 is AckNo=1461
– ACK is cumulative
18
TCP header fields
• Header Length ( 4bits):
– Length of header in 32-bit words
– Note that TCP header has variable length (with minimum
20 bytes)
19
TCP header fields
• Flag bits:
– URG: Urgent pointer is valid
– If the bit is set, the following bytes contain an urgent message in
the range:
SeqNo <= urgent message <= SeqNo+urgent pointer
– ACK: Acknowledgement Number is valid
– PSH: PUSH Flag
– Notification from sender to the receiver that the receiver should
pass all data that it has to the application.
– Normally set by sender when the sender’s buffer is empty
20
TCP header fields
• Flag bits:
– RST: Reset the connection
– The flag causes the receiver to reset the connection
– Receiver of a RST terminates the connection and indicates
higher layer application about the reset
– SYN: Synchronize sequence numbers
– Sent in the first packet when initiating a connection
– FIN: Sender is finished with sending
– Used for closing a connection
– Both sides of a connection must send a FIN
21
TCP header fields
• Window Size:
– Each side of the connection advertises the window size
– Window size is the maximum number of bytes that a
receiver can accept.
– Maximum window size is 216-1= 65535 bytes
• TCP Checksum:
– TCP checksum covers over both TCP header and TCP
data (also covers some parts of the IP header)
• Urgent Pointer:
– Only valid if URG flag is set
22
TCP header fields
• Options:
End of
Options
kind=0
1 byte
NOP
(no operation)
kind=1
1 byte
Maximum
Segment Size
Window Scale
Factor
Timestamp
kind=2
len=4
maximum
segment size
1 byte
1 byte
2 bytes
kind=3
len=3
shift count
1 byte
1 byte
1 byte
kind=8
len=10
timestamp value
timestamp echo reply
1 byte
1 byte
4 bytes
4 bytes
23
TCP header fields
• Options:
– NOP is used to pad TCP header to multiples of 4 bytes
– Maximum Segment Size
– Window Scale Options
» Increases the TCP window from 16 to 32 bits, I.e., the window
size is interpreted differently
» This option can only be used in the SYN segment (first
segment) during connection establishment time
– Timestamp Option
» Can be used for roundtrip measurements
24
Connection Management in TCP
• Opening a TCP Connection
• Closing a TCP Connection
• State Diagram
25
TCP Connection Establishment
• TCP uses a three-way handshake to open a connection:
aida.poly.edu
mng.poly.edu
SYN (Seq
No = x)
ckNo =
A
,
y
=
o
N
q
e
SYN (S
(SeqNo = x
+1, AckNo
x+1)
=y+1)
26
A Closer Look with tcpdump
aida issues
an "telnet mng"
aida.poly.edu
mng.poly.edu
1 aida.poly.edu.1121 > mng.poly.edu.telnet: S 1031880193:1031880193(0)
win 16384 <mss 1460,nop,wscale 0,nop,nop,timestamp>
2 mng.poly.edu.telnet > aida.poly.edu.1121: S 172488586:172488586(0)
ack 1031880194 win 8760 <mss 1460>
3 aida.poly.edu.1121 > mng.poly.edu.telnet: . ack 172488587 win 17520
4 aida.poly.edu.1121 > mng.poly.edu.telnet: P 1031880194:1031880218(24)
ack 172488587 win 17520
5 mng.poly.edu.telnet > aida.poly.edu.1121: P 172488587:172488590(3)
ack 1031880218 win 8736
6 aida.poly.edu.1121 > mng.poly.edu.telnet: P 1031880218:1031880221(3)
ack 172488590 win 17520
27
Three-Way Handshake
aida.poly.edu
mng.poly.edu
S 103188
0193:103
1880193(
win 16384
0)
<mss 146
0, ...>
8586(0)
8
4
2
7
:1
6
8
5
8
8
S 1724
<mss 1460>
0
6
7
8
in
w
4
9
ack 10318801
ack 172488
587 win 175
20
28
TCP Connection Termination
• Each end of the data flow must be shut down independently
(“half-close”)
• If one end is done it sends a FIN segment. The other end
sends ACK.
• Four messages to complete shut down a connection
A
FIN
B
ACK
B can still send to A
FIN
ACK
29
Connection termination with tcpdump
aida issues
an "telnet mng"
aida.poly.edu
mng.poly.edu
1 mng.poly.edu.telnet > aida.poly.edu.1121: F 172488734:172488734(0)
ack 1031880221 win 8733
2 aida.poly.edu.1121 > mng.poly.edu.telnet: . ack 172488735 win 17484
3 aida.poly.edu.1121 > mng.poly.edu.telnet: F 1031880221:1031880221(0)
ack 172488735 win 17520
4 mng.poly.edu.telnet > aida.poly.edu.1121: . ack 1031880222 win 8733
30
TCP Connection Termination
aida.poly.edu
mng.poly.edu
F 172488734:172488734(0)
ack 1031880221 win 8733
. ack 17
2488735
win 174
84
F 10318
80221:1
0318802
ack 1 72
21(0)
488735
win 175
20
222 win
. ack 1031880
8733
31
TCP state diagram
32
TCP States in “Normal” Connection Lifetime
SYN_SENT
(active open)
SYN (SeqNo = x)
y, AckNo
=
o
N
q
e
(S
N
Y
S
=x+1)
LISTEN
(passive open)
SYN_RCVD
(AckNo = y + 1 )
ESTABLISHED
ESTABLISHED
FIN_WAIT_1
(active close)
FIN_WAIT_2
TIME_WAIT
FIN (SeqNo = m)
(AckNo = m+ 1 )
CLOSE_WAIT
(passive close)
FIN (SeqNo = n )
(AckNo =
LAST_ACK
n+1)
CLOSED
33
2MSL Wait State
2MSL Wait State = TIME_WAIT
• When TCP does an active close, and sends the final ACK, the connection
must stay in in the TIME_WAIT state for twice the maximum segment
lifetime.
2MSL= 2 * Maximum Segment Lifetime
A
FIN
ACK
FIN
XACK
B
• Why?
• TCP is given a chance to resent
the final ACK. (Server will timeout
after sending the FIN segment
and resend the FIN)
• The MSL is set to 2 minutes or 1
minute or 30 seconds.
34
Resetting Connections
• Resetting connections is done by setting the RST flag
• When is the RST flag set?
– Connection request arrives and no server process is
waiting on the destination port
– Abort (Terminate) a connection
Causes the receiver to throw away buffered data. Receiver
does not acknowledge the RST segment
35
TCP: Delayed ACKs and Nagle’s
algorithm
36
Interactive and bulk data transfer
TCP applications can be put into the following categories
bulk data transfer
- ftp, mail, http
interactive data transfer
- telnet, rlogin
TCP has heuristics to deal these application types.
For interactive data transfer:
• Try to reduce the number of packets
For bulk data transfer:
• High throughput
37
Telnet session on a local network
Telnet session
from Argon
to Neon
Argon.cs.virginia.edu
Neon.cs.virginia.edu
• This is the output of typing 3 (three) characters :
Time 44.062449: Argon  Neon:
Time 44.063317: Neon  Argon:
Time 44.182705: Argon  Neon:
Push, SeqNo 0:1(1), AckNo 1
Push, SeqNo 1:2(1), AckNo 1
No Data, AckNo 2
Time 48.946471: Argon  Neon:
Time 48.947326: Neon  Argon:
Time 48.982786: Argon  Neon:
Push, SeqNo 1:2(1), AckNo 2
Push, SeqNo 2:3(1), AckNo 2
No Data, AckNo 3
Time 55.116581: Argon  Neon:
Time 55.117497: Neon  Argon:
Time 55.183694: Argon  Neon:
Push, SeqNo 2:3(1) AckNo 3
Push, SeqNo 3:4(1) AckNo 3
No Data, AckNo 4
38
Interactive applications: Telnet
•
Remote terminal applications (e.g., Telnet) send characters
to a server. The server interprets the character and sends
the output at the server to the client.
•
For each character typed, you see three packets:
1. Client  Server: Send typed character
2. Server  Client: Echo of character (or user output) and
acknowledgement for first packet
3. Client  Server: Acknowledgement for second packet
39
Why 3 packets per character?
• We would expect four packets per
character:
character
cter
ACK of chara
c
echo of chara
ter
ACK of echoed character
character
• However, tcpdump shows this
pattern:
ACK and echo
of character
ACK of echoed character
• What has happened?
TCP has delayed the transmission
of an ACK
40
Delayed Acknowledgement
• TCP delays transmission of ACKs for up to 200ms
• The hope is to have data ready in that time frame. Then, the
ACK can be piggybacked with a data segment.
• Delayed ACKs explain why the ACK and the “echo of
character” are sent in the same segment.
41
Telnet session to a distant host
Telnet session
between argon.cs.virginia.edu
and
tenet.cs.berkeley.edu
argon.cs.virginia.edu
tenet.cs.berkeley.edu
• This is the output of typing nine characters :
Time 16.401963:
Time 16.481929:
Argon  Tenet:
Tenet  Argon:
Push, SeqNo 1:2(1), AckNo 2
Push, SeqNo 2:3(1) , AckNo 2
Time 16.482154:
Time 16.559447:
Argon  Tenet:
Tenet  Argon:
Push, SeqNo 2:3(1) , AckNo 3
Push, SeqNo 3:4(1), AckNo 3
Time 16.559684:
Time 16.640508:
Argon  Tenet:
Tenet  Argon:
Push, SeqNo 3:4(1), AckNo 4
Push, SeqNo 4:5(1) AckNo 4
Time 16.640761:
Time 16.728402:
Argon  Tenet:
Tenet  Argon:
Push, SeqNo 4:8(4) AckNo 5
Push, SeqNo 5:9(4) AckNo 8
42
Delayed Acks do not kick in if there are data to
send
• Observation: Transmission
of segments follows a
different pattern, i.e., there
are only two packets per
character typed
char1
r1
+ echo of cha
ACK of char 1
ACK + char2
f
ACK + echo o
char2
• The delayed acknowledgment does not kick in
• The reason is that there is
always data at Argon ready
to sent when the ACK
arrives.
43
Nagle’s Algorithm
• Observation:
– Argon never has multiple unacknowledged segments
outstanding
– There are fewer transmissions than there are
characters.
• Sending one byte per packet is inefficient.
• Solution: Nagle’s Algorithm
Small segments cannot be sent until outstanding data
is acked.
• The algorithm can be disabled, because it could be a
problem to interactive applications such as X window.
44
TCP:
Flow Control
Congestion Control
45
What is Flow/Congestion Control ?
• Flow Control:
Algorithms to prevent that the sender
overruns the receiver buffer
• Congestion Control: Algorithms to prevent that the sender
overloads the network
 Sliding window implements both control mechanisms.
46
TCP Flow Control
47
TCP Flow Control
•
TCP implements sliding window flow control
• Sending acknowledgements is separated from setting
the window size at sender.
•Acknowledgements do not automatically increase the
window size
• Acknowledgements are cumulative
48
Sliding Window Flow Control
• Sliding Window Protocol is performed at the byte level:
Advertised window
1
2
sent and
acknowledged
3
4
5
sent but not
acknowledged
6
7
8
can be sent
USABLE
WINDOW
9
10 11
can't sent
•Here: Sender can transmit sequence numbers 6,7,8.
49
Sliding Window: “Window Opens”
• Acknowledgement is received that enlarges the window to the right
(AckNo = 5, Win=6):
•1
•2
•3
•4
•5
•6
•7
•8
•9 •10 •11
•AckNo = 5, Win = 6
•is received
•1
•2
•3
•4
•5
•6
•7
•8
•9 •10 •11
• A receiver opens a window when TCP buffer empties (meaning that data
is delivered to the application).
50
Window Management in TCP
• The receiver is returning two parameters to the sender
AckNo
window size
(win)
32 bits
16 bits
• The interpretation is:
• I am ready to receive new data with
SeqNo= AckNo, AckNo+1, …., AckNo+Win-1
• Receiver can acknowledge data without opening the window
• Receiver can change the window size without acknowledging
data
51
Sliding Window: Example
Receiver
Buffer
Sender
sends 2K
of data
0
4K
2K SeqNo=0
2K
Sender blocked
Sender
sends 2K
of data
AckNo=2048
Win=2048
2K SeqNo=2
048
4K
Win=0
AckNo=4096
3K
Win
AckNo=4096
=1024
52
TCP Congestion Control
53
TCP Congestion Control
• Keep a sender from congesting the network.
• The sender has two internal parameters:
– Congestion Window (cwnd)
– Slow-start threshhold Value (ssthresh)
• Sliding window size is set to the minimum of (cwnd, receiver
advertised win)
• Congestion control works in two modes:
– slow start (cwnd < ssthresh)
• Probe the available bandwidth
– congestion avoidance (cwnd >= ssthresh)
• Try not to overload the network.
54
Slow Start
• Initial value:
Set cwnd = 1
• Note: Unit is a segment size. TCP actually is based on bytes
and increments by 1 MSS (maximum segment size)
• Modern TCP implementation may set initial cwnd to 2
• Each time an ACK is received by the sender, the
congestion window is increased by 1 segment:
cwnd = cwnd + 1
• If an ACK acknowledges two segments, cwnd is still
increased by only 1 segment.
• Even if ACK acknowledges a segment that is smaller
than MSS bytes long, cwnd is increased by 1.
• Question: how can you accelerate your TCP
download?
55
Slow Start Example
• The congestion
window size grows
very rapidly
– For every ACK, we
increase cwnd by
1 irrespective of
the number of
segments ACK’ed
• TCP slows down the
increase of cwnd
when
cwnd > ssthresh
cwnd = 1
segment 1
t1
ACK for segmen
cwnd = 2
cwnd = 4
segment 2
segment 3
ts 2
ACK for segmen
ts 3
ACK for segmen
segment 4
segment 5
segment 6
ts 4
ACK for segmen
ts 5
ACK for segmen
ts 6
ACK for segmen
cwnd = 7
56
Congestion Avoidance
• Congestion avoidance phase is started if cwnd has reached
the slow-start threshold value
• If cwnd >= ssthresh then each time an ACK is received,
increment cwnd as follows:
• cwnd = cwnd + 1/ cwnd
• So cwnd is increased by one only if all cwnd segments have
been acknowledged.
57
Example of
Slow Start/Congestion Avoidance
Assume that ssthresh = 8
cwnd = 1
cwnd = 2
cwnd = 4
12
10
cwnd = 8
ssthresh
8
6
4
2
cwnd = 9
6
t=
4
t=
2
t=
0
0
t=
Cwnd (in segments)
14
Roundtrip times
cwnd = 10
58
Responses to Congestion
• TCP uses packet loss as congestion signal
• A TCP sender can detect lost packets via:
• Receipt of a duplicate ACK
• Timeout of a retransmission timer
59
Response to Timeout
• TCP interprets a Timeout as a severe congestion signal.
When a timeout occurs, the sender performs:
– cwnd is reset to one:
cwnd = 1
– ssthresh is set to half of the current size of the congestion
window:
ssthressh = cwnd / 2
– and slow-start is entered
60
Reaction to Duplicate ACKs
• Fast retransmit
– Three duplicate ACKs indicate a packet loss
– Retransmit without timeout
• Fast recovery
– Avoid slow start
– Retransmit “lost packet”
– ssthresh = cwnd/2
– cwnd = cwnd+3
– Increment cwnd by one for each additional duplicate ACK
• When ACK arrives that acknowledges “new data” set:
cwnd=ssthresh
enter congestion avoidance
61
Duplicate ACK example
1K SeqNo=0
AckNo=1024
1K SeqNo=1
024
1K SeqNo=2
048
1. duplicate
AckNo=1024
1K SeqNo=3
072
2. duplicate
AckNo=1024
1K SeqNo=4
096
3. duplicate
AckNo=1024
1K SeqNo=1
024
1K SeqNo=5
120
62
Flavors of TCP Congestion Control
• TCP Tahoe (1988, FreeBSD 4.3 Tahoe)
– Slow Start
– Congestion Avoidance
– Fast Retransmit
• TCP Reno (1990, FreeBSD 4.3 Reno)
– Fast Recovery
– Modern TCP implementation
• New Reno (1996)
• SACK (1996)
63
This picture is copied from somewhere
TCP Tahoe
64
SS
This picture is copied from somewhere
TCP Reno (Jacobson 1990)
CA
Fast retransmission/fast recovery
65
TCP III – Retransmission and Timeout
66
Retransmissions in TCP
•
A TCP sender retransmits a segment when it assumes that
the segment has been lost:
1. No ACK has been received and a timeout occurs
2. Multiple ACKs have been received for the same segment
67
Retransmission Timer
•
•
TCP sender maintains one retransmission timer for each
connection
When the timer reaches the retransmission timeout (RTO)
value, the sender retransmits the first segment that has not
been acknowledged
•
The timer is started when
1. When a packet with payload is transmitted and timer is not running
2. When an ACK arrives that acknowledges new data,
3. When a segment is retransmitted
•
The timer is stopped when
– All segments are acknowledged
68
How to set the timer
• Retransmission Timer:
– The setting of the retransmission timer is crucial for good
performance of TCP
– Timeout value too small  results in unnecessary
retransmissions
– Timeout value too large  long waiting time before a
retransmission can be issued
– A problem is that the delays in the network are not fixed
– Therefore, the retransmission timers must be adaptive
69
Setting the value of RTO:
• The RTO value is set based on round-trip time (RTT)
measurements that each TCP performs
Segment 2
Segment 3
ent 2 + 3
egm
ACK for S
Segment
5
RTT #3
• Figure on the right shows three
RTT measurements
t1
en
ACK for Segm
RTT #2
• There is only one measurement
ongoing at any time (i.e.,
measurements do not overlap)
Segment 1
RTT #1
• Each TCP connection measures
the time difference between the
transmission of a segment and
the receipt of the corresponding
ACK
egm
ACK for S
ACK for S
Segme
n
t4
ent 4
egment 5
70
Setting the RTO value
• RTO is calculated based on the RTT measurements
– Uses an exponential moving average to estimate RTT (srtt)
and variance of RTT (rttvar) from
– The influence of past samples decrease exponentially
• The RTT measurements are smoothed by the following
estimators srtt and rttvar:
srttn+1 = a RTT + (1- a ) srttn
rttvarn+1 = b ( | RTT - srttn | ) + (1- b ) rttvarn
RTOn+1 = srttn+1 + 4 rttvarn+1
– The gains are set to a =1/4 and b =1/8
71
Setting the RTO value (cont’d)
• Initial value for RTO:
– Sender should set the initial value of RTO to
RTO0 = 3 seconds
• RTO calculation after first RTT measurements arrived
srtt1 = RTT
rttvar1 = RTT / 2
RTO1 = srtt1 + 4 rttvarn+1
• When a timeout occurs , the RTO value is doubled
RTOn+1 = max ( 2 RTOn, 64) seconds
This is called an exponential backoff
72
Karn’s Algorithm
Timeout !
RTT ?
 RTT measurements is ambiguous in
this case
segme
nt
RTT ?
If an ACK for a retransmitted segment is
received, the sender cannot tell if the
ACK belongs to the original or the
retransmission.
retransm
ission
of segm
ent
ACK
Karn’s Algorithm:
• Don’t update RTT on any segments that have been
retransmitted
73
Summary
• UDP: connectionless, unreliable, datagram service
• TCP: reliable, connection-oriented, byte stream service
– TCP header
– Connection management
– Delayed ACKs and nagle’s algorithm
– TCP flow control
– TCP congestion control
– TCP retransmission and timeout
• References
– TCP/IP illustrated vol. 1, chapter11, 17-24
– RFC793 (Transmission Control Protocol)
– RFC768 (User Datagram Protocol)
– RFC2581 (TCP Congestion control)
– RFC2988 (Computing TCP’s Retransmission Timer)
– RFC3390 (Increasing TCP’s Initial Window)
74