JP2005252638A - Network communication control method and apparatus thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a communication control method and apparatus which detect the condition of a network between communication apparatuses at a high speed to take an adequate measure. <P>SOLUTION: When communication apparatuses are interconnected with a plurality of sessions over a network, session information of the RTT values and/or the congestion window values respectively detected in the plurality of sections are made to be common and updated/set up. Owing to making them common, the communication delay/communication delay recovery in the network is detected at a high speed to quickly deal with it. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a communication control method and apparatus when a plurality of sessions are connected between communication apparatuses.

  As is well known, TCP (Transmission Control Protocol) communication is connection-oriented, and sets a packet retransmission timeout value by measuring a round trip time (RTT). -Flow control to adjust the size can be performed.

  In a conventional communication system, generally, an RTT value and a congestion window value between communication devices are managed on a session basis. For example, Japanese Patent Application Laid-Open No. 7-74780 refers to an “end-to-end” method as a method for adjusting the window pace between a transmission-side device and a transmission-destination device. It has been pointed out that the response is greatly delayed.

  Japanese Patent Application Laid-Open No. 2003-32296 provides a common buffer function for storing a plurality of sessions in a lump, and provides a common service aiming to eliminate a congestion state while ensuring the quality of service (QoS) required for each session. A channel flow control method is disclosed.

JP-A-7-74780 JP 2003-32296 A

  However, when a plurality of sessions are connected between communication apparatuses, packet retransmission timer management and flow control usually vary depending on the network state between the communication apparatuses. If post-control is performed for each session as in the conventional system described above, it is impossible to respond immediately when a network communication delay or communication failure occurs or recovers.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a communication control method and apparatus that can detect a network state between communication apparatuses at high speed and enable appropriate responses.

  According to the present invention, a communication control method in a network system in which communication devices are connected by a plurality of sessions through a network detects a network state detected in each of the plurality of sessions, and a partial session in the plurality of sessions When the change in the network state is detected, the change in the network state is reflected in the session information of all the plurality of sessions.

  The session information may be a round trip time (RTT) value and / or a congestion window value of each section information.

When a new session is connected, the congestion window value CWND _NEW of the new session is
CWND _NEW = (SES _NOW x CWND _NOW ) / (SES _NOW +1)
(Here, SES _NOW is the current number of sessions to the communication device of the other party, and CWND _NOW is the congestion window value currently set.)
And the new congestion window value CWND _NEW is reflected in the existing session information of the plurality of sessions.

Furthermore, the congestion window value CWND _QOS X _NEW used in one session of service is
_{CWND QOS X NEW = T × POINT} QOS × P S
(Where T is the amount of packets between communication devices, calculated by T = Σ (CWND _NOW × SES _NOW ), and C
WND _NOW-S is the current congestion window value, SES _NOW-S is the current number of sessions, and Σ is the sum of the packet amount for each service (CWND _NOW-S × SES _NOW-S ) for all services,
POINT _QOS is a ratio value per policy, and POINT _QOS = 1 / Σ (P _S × SES _NOW-S )
Where P _S is the service policy value and SES _NOW-S is the number of sessions. )
And the new congestion window value CWND _QOS X _NEW is reflected in the session information of all the plurality of sessions.

  According to the present invention, when there are a plurality of sessions connected to the same counterpart communication device, in order to update / set the session information of the RTT value and / or the congestion window value in common, the communication delay / communication High-speed detection and high-speed response for delay recovery are possible.

  In addition, the new session can use the network resources (communication packet amount) used by the existing session, and communication of the new session can be started at high speed.

  Furthermore, by calculating the congestion window value used in one session for each service, priority communication control can be performed by congestion window control incorporating service quality QoS.

  FIG. 1 is a schematic network configuration diagram conceptually illustrating network communication control according to the first embodiment of the present invention. Here, in order to simplify the description, it is assumed that the communication device 1 and the communication device 2 are connected by the network 3 and a plurality of sessions S1 to SN exist between these devices. Note that the network 3 connects between the communication devices 1 and 2, and a plurality of networks may be interposed. According to the present invention, since the RTT value and the congestion window value of each of the sessions S1 to SN are shared, it is possible to cope with the network state at high speed. Details will be described below.

  FIG. 2 is a block diagram showing a schematic configuration of the network communication control apparatus according to the present embodiment. Here, the case of the communication apparatus 1 in FIG. 1 is illustrated.

  In FIG. 2, the network communication control device includes a communication management unit 101, a session management unit 102 directed to the communication device 2, an RTT monitor that monitors RTT measured in each session, and monitors a packet reception interval and an ACK reception interval monitor. 103, a timer management unit 104 for managing a retransmission timer. In the session management unit 102, the session information units 102.1 to 102... N is provided, and the communication management unit 101 sets / updates the RTT value and the congestion window value in each session information.

  The communication management unit 101 simultaneously performs setting / updating on all corresponding session information (here, S1 to SN) when the communication apparatuses (here, the communication apparatus 2) of the communication destination are the same. This communication control will be described by exemplifying a case where a communication delay occurs in the network 3 and is restored.

  When a communication delay occurs in the network 3 between the communication device 1 and the communication device 2, the time until the data packet is transmitted and the confirmation response packet is received increases. Therefore, the RTT monitor 103 of the communication apparatus 1 detects a situation where the RTT measurement values of several sessions increase uniformly. When the RTT monitor 103 detects an increase in RTT measurement values in these sessions, the communication management unit 101 reflects the increase in the RTT values of all the session information S1 to SN. That is, the RTT values of all sessions connected to the same communication device 2 are updated to increased values. As a result, the RTT value can be updated at high speed for all sessions.

  Thereafter, when the communication delay is recovered, similarly, the RTT measurement value decreases uniformly in the session in which the data packet is transmitted and the acknowledgment packet is received. When the decrease in the RTT value is detected, the communication management unit 101 reflects the decrease in the RTT values of all the session information S1 to SN. In this way, the RTT values of all sessions can be updated at high speed to the RTT values according to the network state.

  Similarly, when a communication delay occurs in the network 3 between the communication device 1 and the communication device 2 and then recovered, the reception interval of the acknowledgment packet received with respect to the transmitted data packet is transmitted along with the occurrence of the delay. It becomes longer than the time, and then recovers to return to the normal length. The ACK reception interval monitor 103 monitors that the change in the reception interval occurs uniformly, whereby the communication management unit 101 updates the congestion window value for all sessions as in the case of RTT. As a result, the congestion window value of all sessions can be updated to a value according to the network state at high speed.

Further, when a new session S (N + 1) to the communication device 2 occurs in a state where a plurality of sessions S1 to SN are connected between the communication devices 1 and 2, congestion of the new session S (N + 1) occurs. The window value CWND _NEW is calculated by the following formula, and the value is reflected in all related session sessions S1 to SN.

CWND _NEW = (SES _NOW x CWND _NOW ) / (SES _NOW +1)
Here, SES _NOW is the current number of sessions to the communication apparatus 2, and CWND _NOW is the congestion window value currently set. Therefore, (SES _NOW × CWND _NOW ) represents the amount of network traffic between the current communication device 1 and the communication device 2.

  The congestion window value of a new session is obtained by the above formula and is reflected in other sessions, so that high-speed communication of a new session is possible without increasing the amount of packets between communication devices. At that time, the congestion window value of the existing session decreases, but the congestion window value of the existing session is a fluctuation value between the limit congestion window and the slow start threshold value. Few.

(Other embodiments)
Next, another embodiment of the present invention will be described. In the present embodiment, priority control communication for controlling a congestion window can be performed by incorporating quality of service (QoS). That is, the congestion window value CWND _QOS X _NEW used in one session of the service is calculated by the following formula, and the value is reflected in all the sessions of the related service.

_{CWND QOS X NEW = T × POINT} QOS × P S
here,
T is the amount of packets between communication devices,
T = Σ (CWND _NOW x SES _NOW )
Calculated by That is, (CWND _NOW-S × SES _NOW-S ) is the packet amount for each service, and is obtained by multiplying the current congestion window value CWND _NOW-S and the current session number SES _NOW-S, and Σ is the result. This represents the total for all services, whereby the packet amount T between communication devices is calculated.

POINT _QOS is a percentage value per policy.
_{POINT QOS = 1 / Σ (P} S × SES NOW-S)
Calculated by P _S is the policy value of the service, SES _NOW-S is the number of sessions, the product of the policy value and the number of sessions is totaled for all services, and the reciprocal of the result is the ratio value per policy.

FIG. 3 is a diagram illustrating an example of a setting value of a congestion window in a network communication control method according to another embodiment of the present invention. Here, there are four types of services, Web, Mail, TELNET, and FTP, and the policy value P _S and the number of sessions SES _NOW-S are shown respectively. In this state, when the packet amount T between communication devices is set to 2000, the new congestion window value of the Web service is calculated as follows.

T = 2000,
POINT _QOS = 1 / (4 × 4 + 3 × 2 + 2 × 1 + 1 × 1) = 0.04
Therefore, CWND _QOS X _NEW = 2000 × 0.04 × 4 = 320.

  This is a value higher than the congestion window value averaged by the number of connected sessions, and therefore it is possible to perform Web communication preferentially. In this way, QoS control using the present invention can be implemented.

1 is a schematic network configuration diagram conceptually illustrating network communication control according to a first embodiment of the present invention. It is a block diagram which shows the schematic structure of the network communication control apparatus by this embodiment. It is a figure which shows an example of the setting value of the congestion window of the network communication control method by other embodiment of this invention.

Explanation of symbols

1, 2 Communication device 3 Network 101 Communication management unit 102 Session management unit 103 RTT monitor / ACK reception interval monitor 104 Timer management unit

Claims (5)

In a communication control method in a network system in which communication devices are connected by a plurality of sessions through a network,
Detecting a network state detected in each of the plurality of sessions;
When a change in the network state is detected in a partial session within the plurality of sessions, the change in the network state is reflected in session information of all the plurality of sessions.
A communication control method characterized by the above.   The communication control method according to claim 1, wherein the session information is a round trip time (RTT) value and / or a congestion window value of each section information. When a new session is connected, the congestion window value CWND _NEW of the new session is
CWND _NEW = (SES _NOW x CWND _NOW ) / (SES _NOW +1)
(Here, SES _NOW is the current number of sessions to the communication device of the other party, and CWND _NOW is the congestion window value currently set.)
Calculated by
The new congestion window value CWND _NEW is reflected in the session information of all the existing multiple sessions.
The communication control method according to claim 2. The congestion window value CWND _QOS X _NEW used in one session of service is
_{CWND QOS X NEW = T × POINT} QOS × P S
(Where T is the amount of packets between communication devices, calculated by T = Σ (CWND _NOW × SES _NOW ), and C
WND _NOW-S is the current congestion window value, SES _NOW-S is the current number of sessions, and Σ is the sum of the packet amount for each service (CWND _NOW-S × SES _NOW-S ) for all services,
POINT _QOS is a ratio value per policy, and POINT _QOS = 1 / Σ (P _S × SES _NOW-S )
Where P _S is the service policy value and SES _NOW-S is the number of sessions. )
Calculated by
The new congestion window value CWND _QOS X _NEW is reflected in the session information of all the plurality of sessions.
The communication control method according to claim 2. In a communication control device in a network system in which communication devices are connected by a plurality of sessions through a network,
Detecting means for detecting a network state detected in each of the plurality of sessions;
When detecting a change in the network state in a partial session within the plurality of sessions, communication management means for reflecting the network state change in the session information of all the plurality of sessions;
A communication control apparatus comprising:

Images