ｽﾗｲﾄﾞﾀｲﾄﾙなし

(1)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(2)

ping telnet

ftp

X

traceroute

tftp bootp smtp

NFS/RPC

TCP

UDP

IP

ICMP

IGMP

(3)

インターネットアーキテクチャ

TCP : Transmission Control Protocol

-• TCP (Transmission Control Protocol) ； end-to-end

– フロー制御

– エラー制御 / 再送制御

– コネクション管理

– セッションの多重化

Network

IP

TCP

Application

IP

Application

TCP

Network Network

(4)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(5)

TCP Header Format

0 7 8

15 16

23 24

31 source port identifier

destination port identifier

sequence number

ACK number

0 20 Bytes

Offset(4)

Rsrvd(6)

control bits

UR AK PH R T SY FN

window size

Urgent Pointer

checksum

Option

1

2

3

4

5

6

(6)

TCP Header Format

UR

： Urgent Pointer Field Significant (URG)

AK

： Acknowledgement Field Significant (ACK)

PH

： Push Function

RT

： Reset the Connection

SY

： Synchronize Sequence Numbers (SYN)

FN

： No More Data From Sender (FIN)

(7)

TCP Features

・

_“Stream”

_{Oriented Data Transmission}

→ Connection確立(Three-way-handshake)

・ Connection (“Stream”) Identifier ＝

“Socket”

{dst_IP_addr, dst_port, src_IP_addr, src_port}

・

“Sequence Number”

; 32 bits

→ バイト番号： 0 − (2^32-1)

→ 2^32 でSequence NumberがWrapされる

・

“Full-Duplex”

での通信

・ Acknowledgement (ACK) ;

→ 次に受信すべきバイト番号(SN)の通知

・エラー回復：セグメント再送(Segment retransmission)

by

Time-out, Dupilicated-ACK

・

を用いたデータ転送制御

(8)

TCP Port Allocation （RFC1700)

1. Well-Known Ports

;

0 - 1,023

2. Registered Ports

; 1,024 - 49,151

3. Dynamic and/or Private Ports ; 49,152 - 65,535

TCP Well-Known Ports

Port Number Keyword

Application

5 rje

Remote Job Entry

20 ftp-data

File Transfer [Default data]

21 ftp

File Transfer [Control]

23 telnet

Telnet

25 smtp

Simple Management Protocol

39 rlp

Resource Location Protocol

53 domain

Domain

Name

Server

63 whois++

Whois++

67 bootp

Bootstrap Protocol Server

69 tftp

Trivial File Transfer

70 gopher

Gopher

79 finger

Finger

80 http

World Wide Web HTTP

(10)

TCP Well-Known Ports

Port Number Keyword

Application

123 ntp

Network Time Protocol

137 netbios-ns NetBIOS Name

Service

138 netbios-dgm

NetBIOS Datagram Service

139 netbios-ssn

NetBIOS Session Service

179 bgp

Border Gateway Protocol (BGP)

202 at-nbp

AppleTalk Name Binding Protocol

213 ipx

IPX

220 imap3

IMAP3 (Interactive Mail Access Protocol)

396 netware-ip

Novell Netware over IP

540 uucp

uucp daemon

546 dhcpv6-client DHCPv6 Client

547 dhcpv6-server DHCPv6 Server

(11)

TCP Connection確立/開放

Log on the console;

svr4% telnet bsdi discard

#

port=“9” (server discard packet)

Trying 140.252.13.35

Connected to bsdi.

Escape character is ‘^]’.

^]

telnet> quit

Connection closed.

tcpdump output

1 0.0 svr4.1037 > bsdi.discard: S 14155.14155(0) win 4096 <mss 1024> 2 0.024 (0.0024) bsdi.discard > svr4.1037: S 18239.18239(0) ack 14156 win 4096 <mss 1024> 3 0.007 (0.0048) svr4.1037 > bsdi.discard: . ack 18240 win 4096 4 4.155 (4.1482) svr4.1037 > bsdi.discard: F 14156:14156(0)

ack 18240 win 4096

(12)

TCP Connection確立/開放

tcpdump output

1 0.0 svr4.1037 > bsdi.discard: S 14155.14155(0) win 4096 <mss 1024> 2 0.024 (0.0024) bsdi.discard > svr4.1037: S 18239.18239(0) ack 14156 win 4096 <mss 1024> 3 0.007 (0.0048) svr4.1037 > bsdi.discard: . ack 18240 win 4096 4 4.155 (4.1482) svr4.1037 > bsdi.discard: F 14156:14156(0)

ack 18240 win 4096

5 4.158 (0.0013) bsdi.discard > svr4.1037: . ack 14157 win 4096 6 4.159 (0.0014) bsdi.discard > svr4.1037: F 18240.18240(0)

ack 14157 win 4096

7 4.189 (0.0225) svr4.1037 > bsdi.discard: . ack 18241 win 4096

[意味]

source.port > destination.port : flags SN_begin.SN_end(data_size) flags : S = SYN ; Synchronize sequence_number(SN)

F = FIN ; Finish data transmission R = RST ; Reset connection

P = PSH ; push data to receiving process asap . = ; none of above four flags is on

(13)

TCP Connection確立/開放

SYN

14155.14155(0)

SYN

18239.18239(0) ACK 14156 ACK 18240

FIN

14156:14156(0) ACK 18240 ACK 14157

FIN

18240.18240(0) ACK 14157

svr4.1037 (client)

bsdi.discard(server)

segment 1 segment 2 (14155+1) “次に受信すべきSN” segment 3 (18239+1) “次に受信すべきSN” segment 4 segment 5 (14156+1) “次に受信すべきSN” segment 6

(14)

TCP Connection確立/開放

svr4.1037 (client)

SYN (a)

SYN_ACK(a+1,b)

ACK(b+1)

FIN (m,s)

FIN_ACK (m+1,s)

ACK (m+1)

ACK (s+1)

“Active open”

(appli. open :

telnet

)

“open”

“Active Close”

(application close:

quit

)

“Passive Close”

(application close)

“half close”

bsdi.discard(server)

“Passive open”

“open”

EOF to Application

“half close”

(15)

CLOSED

LISTEN

ESTABLISHED

SYN_SENT

SYN_RCVD

CLOSE_WAIT

LAST_ACK

FIN_WAIT_1

CLOSING

appl: passive open

send: <nothing> _{appl: active open} send: SYN

appl: send data send: SYN Send : RST

recvl: SYN

send: SYN, ACK

recv: SYN send: SYN,ACK (simultaneous open) recv: SYN,ACK send: ACK recv: ACK send: <nothing>

appl: close send: FI

N

appl: close send: FIN

recv: FIN

send: ACK _{appl: close} send: FIN recv: ACK send: <nothing> appl: close or timeout recv: ACK send: <nothing> recv: FIN send: ACK recv: ACK send: <nothing> recv: FIN,ACK send: ACK simultaneous close Active open passive open Passive close

(16)

TCP Layer Interfaces

Session

情報

データ

送信

データ

受信

異常終了

指示

正常終了

指示

開始

指示

Send (Service_Type, TTL, 擬似ヘッダ)

Recieve

TCP

IP

send

_receive

OPEN

SEND

RECEIVE

STATUS

ABORT

CLOSE

Session

確立

データ

送信

データ

受信

Session

_廃棄

Session

_開放

(17)

TCP Layer Interfaces

(1) OPEN Call ：

機能 ; コネクション開始の指示

引数 ; Local_port, Destination_socket, Open_Mode(Active/Passive),

[timeout_value]、[Priority], [security], [Options]

戻り値 ; Local_Connection_Name

TCP動作 ; LISTEN(Passive_Open)、 ESTABLISHED(Active_Open)

System calls; - socket(pf, type, protocol)

- bind(socket, localaddr, adddrlen)

- connect(socket, destaddr, addrlen)

(2) SEND Call :

機能；データの送信指示

引数 ; Local_Connection_Name, 送信データバッファアドレス、

送信データバイト数, [PUSH], [URG], [再送タイムアウト値]

戻り値 ; なし

TCP動作 ; データ送信(ESTASBLISHED)

(18)

TCP Layer Interfaces

(3) RECEIVE Call :

機能；データの受信指示

引数 ; Local_Connection_Name, 受信データバッファアドレス、

受信データバイト数

戻り値 ; 受信バイト数, URG, PUSH, [受信バッファアドレス]

TCP動作 ; データ受信格納 (ESTASBLISHED)

System calls ; read(descriptor, buffer, length)

recvfrom(socket, buffer, flags, fromaddr, addlen)

recvmsg(socket, messagestruct, flags)

(4) STATUS Call ：

機能 ; コネクション状態の取得を指示

引数 ; Local_Connection_Name

戻り値 ; Local_Socket, Destination_Socket, Local_Conenction_Name,

受信window_size, 送信window_size, Connection_state,

ACK待ちバッファ数, 未受信バッファ数, URG, Priority, Security,

(19)

TCP Layer Interfaces

(5) ABORT Call :

機能；コネクションの異常終了指示

引数 ; Local_Connection_Name

戻り値 ; なし

TCP動作 ; コネクション廃棄 (CLOSED)

(6) CLOSE Call :

機能；コネクションの正常終了指示

引数 ; Local_Connection_Name

戻り値 ; なし

TCP動作 ; コネクション開放 (CLOSED)

System call ; close(socket)

(20)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(21)

TCP Interactive Data Flow

Default and Basic Procedure

-Telnet client

_{Telnet server}

echo of data byte

data byte

ack of data byte

key-stroke “d”

to telnet server

process “d”

echo from telnet

server process “d”

echo to display

process “d”

(22)

TCP Interactive Data Flow

Delayed ACK

-Telnet client

_{Telnet server}

data byte

ack of data byte

echo of data byte

to telnet server

process “d”

echo from telnet

server process “d”

Aggregate message

→ Delayed ACK

key-stroke “d”

echo to display

process “d”

(23)

TCP Interactive Data Flow

Delayed ACK : PiggyBack

-Telnet client

_{Telnet server}

echo of data byte

+ ack of data byte

de

lay w

indow

de

lay w

i

data byte

ndow

echo from telnet

server process “d”

(24)

TCP Interactive Data Flow

<Client>

<Server>

date¥n (6 bytes) => Sat Feb 6 07:52:17 MST 1993¥n (30 bytes)

1 0.0 bsdi.1023 > svr4.login: P 0:1(1) ack 1 2 0.016497 (0.0165) svr4.login > bsdi.1023: P 1:2(1) ack 1 3 0.139955 (0.1235) bsdi.1023 > svr4.login: . ack 2

4 0.458037 (0.3181) bsdi.1023 > svr4.login: P 1:2(1) ack 2 5 0.474386 (0.0163) svr4.login > bsdi.1023: P 2:3(1) ack 2 6 0.539943 (0.0656) bsdi.1023 > svr4.login: . ack 3

16 1.799841 (0.0599) svr4.login > bsdi.1023: P 7:37(30) ack 5 17 1.940176 (0.1403) bsdi.1023 > svr4.login: . ack 37

“d”

“a”

“t”

“e”

“¥n”

→“CR/LF”

(25)

bsdi.1023

svr4.login

1

3

4

6

7

9

10

12

13

15

D-ACK D-ACK D-ACK D-ACK D-ACK PSH 0:1(1) ack 1 (d) PSH 1:2(1) ack 1 (echo d) ack 2 ack 3 ack 4 ack 5 ack 7 PSH 1:2(1) ack 2 (a) PSH 2:3(1) ack 2 (echo a) PSH 2:3(1) ack 3 (t) PSH 3:4(1) ack 3 (echo t) PSH 3:4(1) ack 4 (e) PSH 4:5(1) ack 4 (echo e) PSH 4:5(1) ack 5 (¥n) PSH 5:7(2) ack 5 (echo CR/LF) PSH 7:37(30) ack 5 (date内容)

2

5

8

11

14

16 Delayed ACKによる

メッセージのAggregate

・ D-ACK ; 200 msec

・ Piggy-back ;

→ echo + ack

・ segment 13 ; 1 byte data

“¥n”

segment 14 ; 2 byte date

“CR/LF”

(26)

TCP Negle Algorithm

(Receiver)

data packet flow

echo & ack packet flow

(e.g., ack_of_”d” & echo “d”)

aggregate

payload

+

: IP (20B)

: TCP(20B)

: Data

“d”

“e”

“

¥n

”

“t”

_“a”

“¥n”

“e”

“t”

“te¥n”

“Large RTT”

(Sender)

(27)

Negle Algorithm

Telnet client

_{Telnet server}

key-stroke

(sn=1) “c”

(sn=2) “a”

(sn=3) “t”

PSH 1:2(1) ack 2

PSH 2:4(2) ack 3

PSH 2:3(1) ack 2

1

3

5 echo；

“c” (sn=2)

“at” (sn=3,4)

2 ack 2 → OK sn=1

“a”+”t” → “at”

PSH 3:5(2) ack 4

4 ack 4 → “sn=4”を期待

(sn=3まで受信)

_{ack 5}

ack 5

→ “sn=5”を期待

(sn=4まで受信)

(28)

Disable Negle Algorithm

Telnet client

_{Telnet server}

PSH 1:2(1) ack 2

F1 key

(sn=1) “ESC”

(sn=2) “[”

(sn=3) “M”

to telnet server

“ESC” (sn=2)

“[” (sn=3)

“M” (sn=4)

1

2

3

5 PSH 2:3(1) ack 2

PSH 3:4(1) ack 2

PSH 5:6(1) ack 4

PSH 2:5(3) ack 3

4 ack 4 → OK 1,2,3

sn=“5” → missing 2,3,4

_{ack 2}

“^[[”

“M”

ack 2 → “^”を期待

“timeout”

PSH 2:6(4) ack 4

ack 6

“^]]M”

₆

ack 2 以上を受信せず ack 2以降を再送

F1 key echo結果：

“^[[M”

7 _{ack 6 → OK 2,3,4,5}

(29)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(30)

TCP Bulk Data Transmission

Sliding Window

-・ Window制御を用いたパケット転送

①Sliding Window (Receiver設定)

②Congestion Window(Sender設定)

(1) ACKなしにwindow数のパケットを転送

(2) ACKのAggregation(ACKパケットの減少)

(3) Receiver側によるwindow幅の制御

(31)

TCP Sliding Window

Offered window

(advertised by receiver)

Unsent window

1

2

3

4

5

6

7

8

9

10 11 …

sent and

ACKed

Can send ASAP

_{Can not send until}

window slides

(32)

TCP Sliding Window

Sent “3” and “4”

Offered window

(advertised by receiver)

Unsent window

1

2

3

4

5

6

7

8

9

10 11 …

Can not send until window slides

Can send ASAP

sent but not ACKed

sent and ACKed

3+window=9

5+window=11

Receive ack “5”

(33)

TCP Sliding Window

Window advertise by receiver

shrink

enlarge

window

closed by

ACK reception

= ACKed SN

Opend by

ACK reception

(=ack+window)

(34)

bsdi.1023

svr4.discard

1

2

3

4

5

6

7

9

8

10

SYN 0:0(0) win4096 <mss1024>

SYN 3:3(0) ack 1 win4096 <mss1024> ack 4 win4096 PSH 1:1025(1024) ack 4 win4096 PSH 1025:2049(1024) ack 4 win4096 ack 2049 win4096 PSH 3073:4097(1024) ack 4 win4096 PSH 2049:3073(1024) ack 4 win4096 ack 3073 win3072 ack 4097 win4096

11

12

13

14

15

16

PSH 4097:5121(1024) ack 4 win4096 ack 6145 win4096 PSH 7169:8193(1024) ack 4 win4096 PSH 5121:6145(1024) ack 4 win4096 ack 8193 win4096 PSH 6145:7169(1024) ack 4 win4096

18

FI 8193:8193(0) ack 4 win4096 ack 8194 win4096

17

19 ・ Window制御

- window = 4096

- mss = 1024

→ 4 segments は、ACK

なしに転送可能。

・ Window Shrink ;

“7”： 4096 → 3072

**(*) aggregate ACK**

- “7” ← “4” & ”5”

- “10” ← “6” & “9”

- “14” ← “11” & “12”

- “16” ← “13” & “15”

(35)

bsdi.1023

svr4.discard

4

5

6

8

9

10

SYN 0:0(0) win4096 <mss1024>

SYN 3:3(0) ack 1 win4096 <mss1024> ack 4 win4096 PSH 1:1025(1024) ack 4 win4096 PSH 1025:2049(1024) ack 4 win4096 ack 4097 win 0 PSH 2049:3073(1024) ack 4 win4096 ack 4097 win 4096

10

12

13

14

16

PSH 4097:5121(1024) ack 4 win4096 ack 8193 win 0 PSH 5121:6145(1024) ack 4 win4096 PSH 6145:7169(1024) ack 4 win4096 PSH 3073:4097(1024) ack 4 win4096

FIN PSH 7169:8193(1024) ack 4 win4096

ack 8193 win 4096

11

15

1

2

3 ・ Window shrink

- “8” : 4096 → 0

- “14” : 4096 → 0

・ Window enlarge

(= window update)

- “9” : 0 → 4096

- “15” : 0→ 4096

**(*) segment “13” :**

(36)

TCP Congestion Window

Offered window

(advertised by receiver)

Unsent window

1

2

3

4

5

6

7

8

9

10 11 …

Congestion window

(“cwnd”=1 )

sent and

ACKed

Shall not send ASAP

_{Can not send until}

window slides

(37)

TCP Congestion Window

Sent “3”

Offered window

(advertised by receiver)

Unsent window

1

2

3

4

5

6

7

8

9

10 11 …

Can not send until window slides

Shall not send ASAP

Shall send without

ACK ASAP;

cwnd=2 (cwnd

←

**cwnd*2)**

sent and ACKed

4+window=10

(38)

TCP Congestion Window

・

_{Slow Start Policy}

(cwnd ; exponential increase)

cwnd = 1 ;

for (

セグメント転送

₎

{

for (not congestion)

{

if (

セグメント転送

_ACK

受信)

{ cwnd = cnwd +1 }

cwnd = 1

}

**(*)注意 : Congestion Avoidance では若干異なる。**

(39)

TCP Congestion Window

advertised_window

_congestion

cwnd

time

< Congestion

経験の場合 >

< Congestion

なしの場合 >

(40)

TCP Congestion Window(1)

1

1 [受]

[受]

[送]

1

(41)

TCP Congestion Window(2)

2

3

2

3 [受]

[受]

[送]

2

3

2

3

2

3

2

3

(42)

TCP Congestion Window(3)

4 5 4

7 6 5

4 4 5

4 5 6

4 5 6 7

7 6

7 [受]

[受]

[送]

6 5 4

7 6 5 4

(43)

TCP Congestion Window(4)

9 8

8 5 6 7

12 11 10 9

13 12 11 10

14 13 12 11

15 14 13 12

8 8 9

8 9 10

[受]

[送]

10 9 8

6 7

7 11 10 9 8

(44)

bsdi.1029

svr4.discard

5

8

11

12

14

15

16

17

ack 513 win 8192 ack 3073 win 8192 FIN PSH 3585:4097(512) ack 1 win4096

ack 3585 win 8192 513:1025(512) ack1 win4096 1025:1537(512) ack1 win4096 ack 1025 win 8192 1537:2049(512) ack1 win4096 2049:2561(512) ack1 win4096 ack 1537 win 8192 2561:3073(512) ack1 win4096 3037:3585(512) ack1 win4096 ack 2049 win 8192 ack 2561 win 8192 ack 4098 win 7680

2

1:513(512) ack1 win4096

cwnd =1

₁

cwnd =2

3

4 cwnd =3

6

7 cwnd =4

9

10 cwnd =5

cwnd =6

13

(45)

TCP Congestion Window

1． Advertised Window by Receiver

2． Congestion Window (cwnd ) defined by sender

**(*) ローカルに決めることが可能な値**

最適なWindow(Advertised window)サイズ

(1) 廃棄が起こらないくらいの大きさ

→ Congestion Avoidance

(2) 最適パイプライン転送

→ RTT x 帯域幅

(46)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(47)

TCP Data Retransmission

(1) Expire of Retransmission Timeout

(RTO) Value

- RTO calculation using RTT

- Exponential Back-off (Max. 64 sec.)

(2) Reception of Duplicated ACK

- Fast Retransmission / Fast Recovery

(3) Congestion Window (cwnd ) Control

(48)

bsdi.1023

svr4.discard

RTO Expired Retransmission

SYN 3:3(0) ack 1 win4096 <mss1024>

ack 4 win4096 PSH 1:15(14) ack 4 win4096 PSH 15:23(8) ack 4 win4096 ack 15 win 4096 PSH 15:23(8) ack 4 win4096 PSH 15:23(8) ack 4 win4096 PSH 15:23(8) ack 4 win4096 PSH 15:23(8) ack 4 win4096

6

7

8

9

17

SYN 0:0(0) win4096 <mss1024>

1

2

3 再送トライ (RTO; 再送タイマ)

RTO = 1.5 sec /* 変更可能*/

for ( 9 minutes)

{

if ( RTO expired)

{

retransmission;

RTO=RTO x 2;

RTO=min{64sec, RTO};

}

end /* 諦める */

4

5 再送間隔

1.5 sec

3 sec

6 sec

64 sec

(49)

RTO Expired Retransmission

1

0.0 bsdi.1029 > svr4.discard: S 1:1(0)

win 4096 <mss 1024>

2 0.0048 ( 0.0048) svr4.discard > bsdi.1029: S 4:4(0) ack 2

3 0.0064 ( 0.0016) bsdi.1029 > svr4.discard: . ack 5

4 6.1022 ( 6.0958) bsdi.1029 > svr4.discard: P 1:15(14) ack 5

5 6.2594 ( 0.1571) svr4.discard > bsdi.1029: . ack 15 win

6 24.4801 (18.2207) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

7 25.4937 ( 1.0136) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

8 28.4937 ( 3.0001) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

9 34.4937 ( 6.0002) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

10 46.4844 (11.9905) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

11 70.4851 (24.0007) bsdi.1029 > svr4.disacrd: P 15:23(8) ack 5

12 118.4864 (48.0013) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

13 182.4881 (64.0018) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

14 246.4899 (64.0018) bsdi.1029 > svr4.discard: P 15:28(8) ack 5

15 310.4910 (63.9917) bsdi.1029 > svr4.discard: P 15:23(8) ack 5

(50)

RTO Expired Retransmission

RTO Exponential BackOff

-64 sec

32 sec

16 sec

面積(S)

If (S ≦ 9 minutes)

{

continue;

}

else

{

abort;

}

RTO

48 sec

0 1 2 3 4 5 6 7 8 9 10 11

(51)

Timeout and Retransmission

RTO (Retransmission TimeOut)

(1) 古い計算方法 ; RTTの変動に弱い

- RTO = 2 x RTT

where RTT = α･RTT

p

+ (1-α)･RTT

M

= 0.9xRTT

p

+ 0.1xRTT

M

(2) 最近の計算方法 ; 平均偏差の利用

- RTO = 平均RTT + 4 x 平均偏差

(52)

Retransmission by Duplicated ACK

(2) Reception of Duplicated ACK

- Fast Retransmission / Fast Recovery

Segment廃棄特性 ;

→ “single (or few) segment(S)” あるい

は連続多数。

→ 未ACKの同一ACK Segmentsを

複数(3回)受信したら、再送。

(53)

Fast Retransmission by Duplicated ACK

6401:6657(256) ack1

6657:6913(256) ack1

6913:7169(256) ack1

8193:8449(256) ack1

6657:6913(256) ack1

“Fast Retransmission”

ack 5889

ack 6145

ack 6401

7169:7425(256) ack1

ack 6657

7425:7681(256) ack1

_{ack 6657}

_①

7681:7937(256) ack1

_{ack 6657}

_②

7937:8193

（

256) ack1

ack 6657

③

ack 6657

ack 8449 win5888

8449:8705(256) ack1

(54)

Congestion Window Control

[目的]

cwndの大きな振動を防ぎ、

適切なcwndで運用する

[1] cwndの制御

(i) ssthresh以下のcwndサイズ

→ Exponential increase

(slow start)

(ii) ssthresh以上のcwndサイズ

→ Liner increase

(congestion avoidance)

[2] ssthreshの制御

(i) Timeout ； goto “1”

cwnd=1;

ssthresh=65KB;

for ()

{

if (“Timeout”)

{ cwnd=1;

ssthresh = cwnd/2;

}

if (“duplicated ACK”)

{ ssthresh=cwnd / 2;

cwnd=ssthresh;

}

if (cwnd

≦

ssthresh)

{ slow_start;

/* exponential */

}

else

{ congestion_avoidance;

/* liner */

}

(55)

Congestion Window Control (続)

・ ICMP 制御メッセージ

(1) ICMP Source Quench

→ cwnd = 1 ;

ssthresh = as is ;

(2) Host unreachable

(56)

Congestion

avoidance

slow-start

Congestion

avoidance

Congestion

avoidance

slow-start

“ssthresh”

(cwnd_

1)

/ 2

(c

wnd_3

) /

2 cwdn_1

Target

cnwd

cwdn_3

cwnd

(57)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

(58)

TCP Persist Timer

[機能]

Advertised window size = “0” でも

1 Byte のデータは送信できるように

する。

[目的]

window=“0”時のデッドロックの回避。

受信側プロセスの生存確認

[Timer値制御]

(59)

TCP Persist Timer

1 bsdi.1027 > svr4.5555: P 1:1025(1024) ack 1 win 4906 2 ( 0.1920) svr4.5555 > bsdi.1027: . ack 1025 win 4906

3 ( 0.0050) bsdi.1027 > svr4.5555: . 1025:2049(1024) ack 1 win 4096 4 ( 0.0034) bsdi.1027 > svr4.5555: . 2049:3073(1024) ack 1 win 4096 5 ( 0.0072) svr4.5555 > bsdi.1027: . ack 3073 win 4096

6 ( 0.0052) bsdi.1027 > svr4.5555: . 3073:4097(1024) ack 1 win 4096 7 ( 0.0034) bsdi.1027 > svr4.5555: P 4097:5121(1024) ack 1 win 4096 8 ( 0.0039) bsdi.1027 > svr4.5555: P 5121:6145(1024) ack 1 win 4096 9 ( 0.0079) svr4.5555 > bsdi.1027: . ack 5121 win 4096

10 ( 0.0051) bsdi.1027 > svr4.5555: P 6145:7169(1024) ack 1 win 4096 11 ( 0.0040) bsdi.1027 > svr4.5555: P 7169:8193(1024) ack 1 win 4096 12 ( 0.0039) bsdi.1027 > svr4.5555: P 8193:9217(1027) ack 1 win 4096 13 ( 0.1612) svr4.5555 > bsdi.1027: . ack 9217 win 0

14 ( 4.9494) bsdi.1027 > svr4.5555: . 9217:9218(1) ack 1 win 4096 15 ( 0.0040) svr4.5555 > bsdi.1027: . ack 9217 win 0

(60)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(61)

Keep Alive Timer

[目的] ；プロセスダウンへの対応

**(*) TCPコネクションの終了条件**

(i) 明示的終了 (FIN segment)

(ii) プロセスのダウン

[機能] ； TCPコネクションの生存確認

- 2時間ごとにprobe

- 10回(75秒間隔でprobe送信)応答なし

(62)

TCP Keepalive Timer

1 bsdi.1055 > svr4.echo: P 1:14(13) ack 1 2 ( 0.0061) svr4.echo > bsdi.1055: P 1:14(13) ack 14 3 ( 0.0087) bsdi.1055 > svr4.echo: . ack 14

4 (7199.8797) arp who-has svr4 tell bsdi

5 ( 0.0021) arp reply svr4 is-at 0:0:c0:c2:9b:26 6 ( 0.0009) bsdi.1055 > svr4.echo: . ack 14

7 ( 0.0041) svr4.echo > bsdi.1055: . ack 14 8 (7200.1545) arp who-has svr4 tell bsdi

9 ( 0.0021) arp reply svr4 is-at 0:0:c0:c2:9b:26 10 ( 0.0009) bsdi.1055 > svr4.echo: . ack 14

11 ( 0.0040) svr4.echo > bsdi.1055: . ack 14

12 (7200.1769) arp who-has svr4 tell bsdi ① 13 ( 75.0021) arp who-has svr4 tell bsdi ② 14 ( 75.0020) arp who-has svr4 tell bsdi ③ 15 ( 75.0021) arp who-has svr4 tell bsdi ④ 16 ( 75.1123) arp who-has svr4 tell bsdi ⑤ 17 ( 75.0021) arp who-has svr4 tell bsdi ⑥ 18 ( 75.0020) arp who-has svr4 tell bsdi ⑦ 19 ( 74.9920) arp who-has svr4 tell bsdi ⑧ 20 ( 75.0018) arp who-has svr4 tell bsdi ⑨

(63)

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(64)

トランスポートレイヤ技術

その他・将来

-(1) Silly Window Syndrome

(2) MTU Discovery

(3) Window scaling for long fat-pipe

(4) T/TCP (Transaction TCP)

(5) Rate Control

(65)

Path MTU Discovery

[目的]

経路上でフラグメントされない最大のセグメン

トサイズ(Path MTU)の検索

[方法]

ICMPエラーメッセージの利用

(DF; Don’t Fragment オプション)

［頻度］

(66)

1 (0.0 ) solaris.33016 > slip.discard: S 1:1(0) win 8760 <mss 1460> (DF) 2 (0.1016) slip.discard > solaris.33016: S 1:1(0) ack 1

win 4096 <mss 512>

3 (0.5290) solaris.33016 > slip.discard: P 1:513(512) ack 1 win 4096 <mss 512>

4 (0.0038) bsdi > solaris: icmp: slip unreachable - need to frag, mtu = 296 (DF)

5 (0.0259) solaris.33016 > slip.discard: F 513:513(0) ack 1 win 9216 (DF)

6 (0.0923) slip.discard > solaris.33016: . ack 1 win 4096 7 (0.3577) solaris.33016 > slip.discard: P 1:257(256) ack 1

win 9216 (DF)

8 (0.3290) slip.doscard > solaris.33016: . ack 257 win 3840 9 (0.3308) solaris.33016 > slip.discard: FP 257:513(256) ack 1

win 9216 (DF) 10 (0.3208) slip.discard > solaris.33016: . ack win 3840

11 (0.0422) slip.discard > solaris.33016: F 1:1(0) ack 514 win4096 12 (0.1719) slip.discard > splaris.33016: . ack 2 win 9216 (DF)

<mss 1460>

<mss 296>

(67)

Window Scaling for Long Fat Pipe

RFC1323

-Network Bandwidth（bps）

RTT（ms）

BWxRTT（B）

Ethernet

10.000 M

3 3,750

T1（大陸間）

1.544 M

60 11,580

T1（衛星）

1,544 M

500 96,500

T3（大陸間）

45,000 M

60 337,500

OC12（大陸間）

2,400,000 M

60 7,500,000

・ Max. Window Size ； 2^(16) Bytes = 64KB

→ Window Scaling ; “wscale”

(68)

Window Scaling for Long Fat Pipe

1 vangogh.4107 > bsdi.echo: S 1:1(0) win 65535

2 ( 0.0031) bsdi.echo > vangogh.4107: S 1:1(0) ack 1 win 4906 <mss 512> 3 ( 0.2972) vangogh.4107 > bsdi.echo: . ack 1 win 65535

4 (16.6198) vangogh.4107 > bsdi.echo: P 1:14(13) ack 1 win 65535 5 ( 0.0030) bsdi.echo > vangogh.4107: P 1:14(13) ack 14 win 4096 6 ( 0.2971) vangogh.4107 > bsdi.echo: . ack 14 win 65535

7 ( 9.4202) vangogh.4107 > bsdi.echo: F 14:14(0) ack 14 win 65535 8 ( 0.0024) bsdi.echo > vangogh.4107: . ack 15 win 4096

9 ( 0.0013) bsdi.echo > vangogh.4107: F 14:14(0) ack 15 win 4096 10 ( 0.2363) vangogh.4107 > bsdi.echo: . ack 15 win 65535

11 (17.5200) vangogh.4107 > bsdi.echo: S 1:1(0) win 65535

12 ( 0.0031) bsdi.echo > vangogh.4107: S 1:1(0) ack 1 win 4096 <mss 512> 13 ( 0.2967) vangogh.4107 > bsdi.echo: . ack 1 win 65535

nop; no operation

(69)

RFC 1379 ; T/TCP

Transaction TCP

-[目的]

TCPコネクションの確立・開放手続きの

速度アップ

[方法]

・ CC (Connection Count) Option

・ SYNへのPiggy-back ; “half-synchronization”

(1) SYN, Data, FIN, CC

(2) SYN, SYN-ACK, Data, FIN, FIN-ACK,

CC, CC-Echo

(70)

RFC 1379 ; T/TCP

Server

Client

SYN (a)

SYN_ACK(a+1,b)

ACK(b+1)

FIN (m,s)

FIN_ACK (m+1,s)

ACK (m+1)

ACK (s+1)

Data (a+2)

Data_ACK(a+2,b+1)

Server

Client

SYN,Data,FIN,CC

SYN,S-ack,Data,

F,F-ack

FIN-ACK

9 セグメント

→ 3 セグメント

(71)

TCPレート制御

Destination

Node 1

Source

Node 1

Window

制御

Rate

制御

Shaped Transmission with Window Line Speed

(72)

TCPレート制御

Window

制御

Line Speed with Window

Wind

ow

帯

パケット廃棄平均転送速度パケット廃棄ピーク転送速度パケット廃棄

Rate

制御

Shaped Transmission with Window

Wind

ow

帯

平均転送速度ピーク転送速度

(73)

ECN(Explicit Congestion Notification)制御

TOS for Differentiated Service

- PHB(Per-Hop-Behavior)

- CU(Currently Unused)

=> for ECN(Explicit Congestion Notification) ?

0 1 2 3 4 5 6 7

TOSフィールド:

PHB: 000000 DE (Default Service)

101110 EF (Expedited Forwarding)

Others AF (Assured Forwarding)

xxxxx0 Standard Purpose

(74)

Explicit Congestion Notification (ECN)制御

Destination

Node 1

Source

Node 1

(1) ECN=00

(4) ECN=10

(6) ECN=11

(5) ECN=11

Reduce Speed

(2) ECN=01

(3) ECN=01

(7) ECN=10

(8) ECN=11

(9) ECN=11

Congestion Node

ｽﾗｲﾄﾞ ﾀｲﾄﾙなし

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

ping telnet

ftp

X

traceroute

tftp bootp smtp

NFS/RPC

TCP

UDP

IP

ICMP

IGMP

インターネットアーキテクチャ

TCP : Transmission Control Protocol

-• TCP (Transmission Control Protocol) ； end-to-end

– フロー制御

– エラー制御 / 再送制御

– コネクション管理

– セッションの多重化

IP

TCP

Application

IP

IP

Application

TCP

トランスポートレイヤ技術

TCP; Transmission Control Protocol

-(1) TCPの基本動作

(2) TCP Interactive Data Flow

(3) TCP Bulk Data Flow

(4) TCP Data Retransmission

(5) TCP Persist Timer

(6) TCP Keep Alive Timer

(7) その他・将来

TCP Header Format

0

7 8

15 16

23 24

31

source port identifier

destination port identifier

sequence number

ACK number

0

20 Bytes

Offset(4)

Rsrvd(6)

window size

Urgent Pointer

checksum

Option

1

2

3

4

5

6

TCP Header Format

UR

： Urgent Pointer Field Significant (URG)

AK

： Acknowledgement Field Significant (ACK)

PH

： Push Function

RT

： Reset the Connection

SY

： Synchronize Sequence Numbers (SYN)

FN

ｽﾗｲﾄﾞﾀｲﾄﾙなし

_“Stream”

_{Oriented Data Transmission}

→ バイト番号： 0 − (2^32-1)

・エラー回復：セグメント再送(Segment retransmission)