CN105900385B - Load distribution in data communication network - Google Patents
Load distribution in data communication network Download PDFInfo
- Publication number
- CN105900385B CN105900385B CN201480036848.8A CN201480036848A CN105900385B CN 105900385 B CN105900385 B CN 105900385B CN 201480036848 A CN201480036848 A CN 201480036848A CN 105900385 B CN105900385 B CN 105900385B
- Authority
- CN
- China
- Prior art keywords
- connection path
- data
- connection
- communication network
- path
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
Abstract
A kind of method for load to be assigned to numerous connection paths (130A, 130B, 130C, 130D) in data communication network is provided for data communication network, the data communication network includes at least two connection paths, that is, the first connection path (130A) and the second connection path (130B) that interconnect first device (T1) and second device (T2).It the described method comprises the following steps: data flow of the monitoring (S11) on the first connection path (130A) and on second connection path (130B)；Determine the data transport behavior of (S12) in the transport layer (114,124) of the first connection path (130A) and described second connection path (130B)；Determine first data throughout of (S13) described first connection path (130A) and the second data throughout of second connection path (130B)；The first connection path (130A) and second connection path (130B) are arrived into load distribution (S14) based on first data throughout and second data throughout.
Description
Technical field
The present invention relates to the load distribution in data communication network.In particular it relates to a kind of in data
Method and one kind that load is assigned to numerous connection paths are used to be assigned to load in data communication network in communication network
The load-balancing device of numerous connection paths.
Background technique
The terminal installation that data communication network allows attachment to data communication network is exchanged by the connection path of data communication network
Data.Terminal installation may, for example, be such as client (usually request service or data) or server (usually will service and data
Client is provided) etc. computers, the computer be connected to such as internet or any other public or private data network
Data communication network, and for by data communication network exchanging data.It is filled in general, two terminal installations can be referred to as request end
Set (client) and service terminal device (server).It is filled from first terminal device to second terminal in general, data communication network is provided
The multiple and different paths set.These different paths are referred to as the connection path between first terminal device and second terminal device.
Data communication network may include multiple wide area networks (wide area networks, WAN), and the wide area network can be
Part, region or the logic subsystem of data communication network.There may be wherein need to fill a terminal using multiple wide-area network links
Set the service condition for being connected to data communication network, that is, the data network access node of for example more wan routers passes through individual link
It is connected to more than one WAN of data communication network.
This configures the assessment that may need business of the load balancing in multiple wide-area network links and the quality to wide-area network link, with
Just it is prioritized utilization rate and makes it possible to allow link failure detection.
Summary of the invention
It is an object of the present invention to improve the mechanism of load distribution in data communication network.
It has realized that the standard for assessing link quality may be complicated and be difficult to set up, because data are logical
Letter net connection may reach different terminal installations.Different services can be reached by different routings by data communication network
End device, and the Operations, Administration and Maintenance (operation, administration, and maintenance, OAM) of standard
Technology may act only on dedicated point-to-point connection.
The conventional mechanism of quality for detecting the connection path in data communication network be determining service terminal device whether
It is available, that is, the connection of this service terminal device whether can be arrived by giving connection path foundation.This can be based on by asking
End device is asked to send short request (such as by sending ping message or by sending DNS request) to service terminal device to be used for really
The fixed connection whether provided through specific connection path.These mechanism may be time-consuming and slow process, and the process can
The complete loss by the connection of connection path can only be detected and cannot be assessed with regard to such as available bandwidth or for the response time
The quality of link.
According to the first aspect of the invention, it provides a kind of for load to be assigned to numerous connections in data communication network
The method in path, wherein the data communication network includes at least two connection paths, that is, interconnect first device and second device
First connection path and the second connection path.The described method comprises the following steps: monitoring is on first connection path and the
Data flow on two connection paths；Determine that the data in first connection path and the second connection path transport layer transmit row
For；Determine the first data throughout of first connection path and the second data throughout of the second connection path；Based on the first number
Load is assigned to first connection path and the second connection path according to handling capacity and the second data throughout.
The quality of the one or more connection paths of assessment is solved the problems, such as according to the method for the present invention, so as to these connections
The load balancing in path can be carried out based on this assessment.The method can be used for having any number of connection path and terminal
The data communication network of device.Hereinafter, the method is described with reference to the first and second connection paths.It will therefore be noted that
The details that below with reference to the first and second connection paths provide is similarly applicable for any required number in data communication network
Purpose connection path.
Data transport behavior is monitored in the transport layer of the first and second connection paths.In data communication network, especially
In a computer network, transport layer provides end-to-end or host for the application in the layered architecture of networking component and agreement
To main-machine communication service (connection between first terminal device and second terminal device).Transport layer is provided for example towards connection
The service such as data flow support, reliability, flow control and multiplexing.Therefore, the method is based on for example passing through the first link road
Diameter and/or multiple end to end connection by the second connection path or other connection path from first device to second device
Load in data communication network is assigned to connection path by quality.
In other words, the described method comprises the following steps: assessment is specifically connected in one or more of data communication network
The quality in path, and load is assigned to one or more connection paths by the quality based on the connection path assessed.
The quality of assessment connection path may include one or more of following movement and the quality can pass through phase
That answers characterizes through measurement parameter.It can determine through each of the available connection path between the first and second devices
Data throughout from first device to second device or vice versa.These may include for example determine available bandwidth or it is per when
The step of data volume of gap.
Load, which is assigned to connection path, may comprise steps of: load balancing is on the first and second connection paths
Business, priorization can use the utilization rate of connection path.The method needs and using the letter than being provided by only link failure detection
Cease more information.However, the step of the method may include the failures of detection connection path, that is, detection is first and the
The step of fracture or interruption of connection path between two devices.As provided above with methodologies described hereinafter to end-to-end
The reliable measurements of link-quality, without being only about the whether available information of connection path.It is such as above and described below
Method especially provides the assessment of the availability to one or more connection paths.
The possible form of implementation of first according to the first aspect of the invention monitors on first connection path and second
The step of data flow on connection path, includes the steps that the input traffic for monitoring first device.
Therefore, the method can the only upper implementation of one in terminal installation.Input traffic at first device
It is corresponding with the output stream of second device.When monitoring the input traffic of first device, can determine received in institute
Time between data packet and the quality that therefore can assess connection path.Therefore, even if only implementing or holding at a device
Row the method step, can also assess the quality in end to end connection path.This, which can be reduced, implements side described herein
The cost of method.
According to the first aspect of the invention itself or the first form of implementation according to first aspect the second possible reality
Form is applied, monitoring is including monitoring first device on first connection path and in the step of data flow on the second connection path
The step of output stream.
When monitoring output at first device and input traffic, can carry out to along arriving one of second device
The assessment of the end to end connection quality of connection path, because outputting and inputting data flow respectively and by second device and first device
The data packet of transmission is corresponding.This embodiment may need data packet or data flow being assigned to connection path to assess difference
The quality of connection path.
It can be monitored at client or at server and output and input data flow.Input data can be especially by
One or the confirmation packet that sends of second device, the reception of the confirmation packet confirmation to the packet from another device.In other words, it supervises
The step of depending on input traffic may include the confirmation packet in analysis end to end connection between two.
According to the first aspect of the invention itself or any of the above form of implementation according to first aspect
Three possible forms of implementation, data communication network include that the third connection path for interconnecting first device and 3rd device and the 4th connect
Connect path.As above with reference to described by the first device and second device interconnected using first connection path and the second connection path
The step of be respectively suitable for third connection path and the 4th connection path.
In other words, for load distribution method can be applied to respectively with end to end connection between two
Corresponding numerous connection path.Data communication network can especially be configured for making n device interconnecting.Therefore, it counts herein
According to can establish a specific end to end connection of n × (n-1) in communication network in total.One specific device n1 and a specific device
N2 is referred to as a device pair.Each device from n device by numerous connection paths to can be interconnected.In other words
It says, each end to end connection can be established by more than one connection path.It, can be according to first aspect for each pair of device
Method or embodiment execute the load to numerous connection paths of related device pair and distribute.Therefore, for being filled at two
Each end to end connection between setting determines an optimal connection path.
4th possible form of implementation of the third form of implementation according to the present invention according to first aspect, the method into
The numerous connection paths for coming from first device are grouped by one step the following steps are included: depending on connected device, so that will
Numerous connection paths from first device to second device are grouped into first group and by the crowds from first device to 3rd device
More output connection paths are grouped into second group；And it will first aspect relevant to first connection path and the second connection path
The step of be separately to the connection path of first group and second group.
This embodiment makes it possible to determine an optimal connection path for every n1 to n2 end to end connection, allow to
Pass through the net guided data flow from first device to second device of data communication respectively.
Between two devices n1 and n2, more than one data flow can establish.For example, described device can be with more
The computer of a application program.The one or more application program run on device n1 may need to run on device n2
One or more application program carry out data exchange.All data flows of one device n1 can classify according to destination, make
There must be the data flow of identical destination device n2 in an identical group.It can individually be executed for each of these groups negative
Carry it is balanced and to the assessment of the quality of data link to identify the optimal connection between related two devices n1 and n2
Path.
According to the first aspect of the invention itself or any of the above form of implementation according to first aspect
Five possible forms of implementation determine the data transport behavior in first connection path and the second connection path transport layer
Step includes the steps that the TCP session behavior of determining first connection path and the second connection path.
TCP is the transmission control protocol of TCP/IP model.In general, transport layer embodiment is included in data network embodiment party
TCP/IP model and open system interconnection (Open Systems Interconnection, OSI) model of case and general networking
In the two.TCP is known transport protocol.It is used for as above with reference to described in end to end connection between two
Connection-oriented transmission.For the purpose of this description, the end to end connection of TCP session identification between two.
With TCP on the contrary, connectionless User Datagram Protocol (User Datagram Protocol, UDP) is for simpler
Single message transmission transmission.Method described herein can be applicable to arrive by the UDP and other ends of data communication network
Hold connection protocol.
One of TCP connection can be determined for each end to end connection, more or all parameters be to determine from one
Optimal connection path of a terminal installation to another terminal installation.Such TCP connection parameter may, for example, be transmitted and/or institute
The sequence number of received data packet.Another parameter can be the number of the data packet by reception device reception and confirmation.In addition, can
Data packet is being sent from first device to second device and is wrapping it from second device reception for the confirmation of this data packet to determine
Between time delay.Can additionally or alternatively determine two from second device that are to receive at first device it is subsequent or
Time delay between continuous data packet.All or part in these parameters can be used for assessing in first device and second device
Between end to end connection data link quality.
6th possible form of implementation of the 5th form of implementation according to the present invention according to first aspect, determines the first company
The step of connecing the TCP session behavior of path and the second connection path includes determining that the window of TCP session at the appointed time section is big
Small behavior.
Window size is the parameter of TCP connection and is needing for before confirming to the received confirmation packet of these data packets
Identify the amount of received data packet.In general, depending on the amount of lost data packet, that window is adjusted between data transfer period is big
It is small.This usually passes through ICP/IP protocol stack and automatically carries out.The change of window size over time can be used for assessing end and arrive
It holds the quality of connection and is accordingly used in determining that the optimal connection path between the first and second devices, the connection path are subsequent
Further data for data access network node are transmitted.Particularly, the stability window size at relatively high level
(that is, send confirmation packet before received data packet number it is relatively high) can indicate connection path good quality and
Corresponding connection path can be conducive to data transmission.
7th possible form of implementation of the 5th or the 6th form of implementation according to the present invention according to first aspect determines
The step of TCP session behavior of first connection path and the second connection path includes determining by received first data of first device
Time delay between frame and the second data frame, wherein the first data frame and the second data frame are the continuous data frames of TCP session.
The time delay between the data packet or continuous data packet that every two is subsequently received can be determined to identify transmission
Temporal variation.However, the time delay in particular, in that at sending device there are continuous output stream in the case where can be with
It is the parameter helped, so that at reception device, it can be assumed that the variation of the time delay between subsequent or continuous data packet
It is the result by the transmission of connection path.
One the 8th possible implementation in the 5th to the 7th form of implementation according to the present invention according to first aspect
Form, the step of determining the TCP session behavior of first connection path and the second connection path include that determination is sent out by first device
The time delay being sent between the request of second device and the response from second device.
It is generally assumed that request and response data packet are sent by data communication network by identical connection path.Cause
This, this parameter can be the reliable indicator of the transmission time for the data packet by specific connection path.
For example, for each connection path in numerous available connection paths, by from first device to the second dress
Before the connection path transmission user data set, second device can be transmitted the request to and determined until receiving response
Time delay.This request can be such as connection and establish message, and responds and can be connection confirmation message.By numerous connection paths
Time delay be compared each other and can choose that connection path with minimal time delay to be used for transmission user data.
As the method with reference to described in time delay between request and response can be applied to it is described herein its
Its parameter.
One the 9th possible implementation in the 5th to the 8th form of implementation according to the present invention according to first aspect
Form, the step of determining the TCP session behavior of first connection path and the second connection path include that determination is received by first device
Continuous data frame sequence number and identify whether to lose a certain frame in connection path.
In general, data packet need not and can not be received by the order for having sent data packet.The transmitted data packet of exchange
Order this result may due to data communication network internal cause and occur.However, receiving order can be connection path
Quality indicator.Connection path with fewer number of packet Order exchange event can be preferred for transmitting number of users
According to.
Preferred connection path can based on one in parameter described herein, some or all determine.It can
The evaluation of connection path is determined so that metric is applied to the parameter.It can be before determining the assessed value of connection path
Factor is applied to the parameter.The factor of parameter can change or can keep constant.
According to the first aspect of the invention itself or any of the above form of implementation according to first aspect
Ten possible forms of implementation are such as further included steps of with methodologies described hereinafter by machine learning algorithm above
It is applied to the monitored data flow of first connection path and the second connection path, to find to connect in first connection path and second
Connect the connectivity problem on path.
By using normal operating functional parameter is established based on the abnormality detection of machine learning, it is inferred in different companies
It connects on path to the state of the connection of specific destination and therefore the state can be used for intelligent load balancing.In this step
In, all or some in parameter as described above can be considered in machine learning algorithm.
The machine learning algorithm can establish link baseline and obtain exception from it.In other words, can learn to connect
The normal behaviour in path, that is, can determine and analyze the range of the parameter of corresponding connection path to have description connection path
Common or normal behaviour each parameter range.Can then be observed between data transfer period the parameter and with it is normal
The deviation of parameter area may be used as the indicator of the changed quality of connection path.It is instruction link road depending on the parameter
The deterioration of the transmission quality of diameter still improves, and may not further use this connection path or using for the connection path may
Increase, i.e., more multiple data stream is sent by this connection path.
11st possible form of implementation of the tenth form of implementation according to the present invention according to first aspect, by load point
The step of being fitted on first connection path and the second connection path is executed based on the machine learning algorithm.
By machine learning algorithm be used for load balancing can provide high flexibility relevant to used connection path and
Can the variation to data communication network total load or area load quickly make a response.
According to the second aspect of the invention, it provides for load to be assigned to numerous connection paths in data communication network
Load-balancing device.Data communication network includes at least two connection paths, that is, make that first device and second device interconnect the
One connection path and the second connection path.Load-balancing device includes connection tracking module, analytical unit and Load Balance Unit.
Connection tracking module is used to monitor the data flow on first connection path and on the second connection path.Analytical unit is for true
The data transport behavior being scheduled in first connection path and the second connection path transport layer.Analytical unit is further used for really
Determine the first data throughout of first connection path and the second data throughout of the second connection path.Load Balance Unit is used for
Load is assigned to first connection path and the second connection path based on the first data throughout and the second data throughout.
Load-balancing device and its component particularly for executing such as above and below with reference to the first aspect of the present invention and
Embodiment is described for that will load the method for being assigned to connection path.
Connection tracking module can be particularly for receiving and analyzing all to output and input number according to the strategy that define in advance
According to packet, the strategy is such as destination, application type, quality of service requirement and priority.Connection tracking module continuously monitors
All data flows and its variation behavior is recorded, as described by parameter of the reference end above to end connection.
Analytical unit can be used the machine learning such as machine learning algorithm and abnormality detection practice come detect connectivity problem and
The variation of the link-quality of connection path.Analytical unit can be further used for dynamically changing the configuration of Load Balance Unit.
The load-balancing decision of Load Balance Unit implementation analytical unit.
Load-balancing device may include the other component such as packet inspection module and storage unit.Packet inspection module
It can be used to implement the fine grit classification for outputting and inputting data packet and data flow.Storage unit can be used for storing about data
The strategy and information of stream, connection path and its statistical data or parameter.
Itself first possible form of implementation according to the second aspect of the invention, load-balancing device further comprise using
In the first interface for being connected to first device and second interface for being connected to data communication network.Load Balance Unit is for leading to
It crosses second interface and load is assigned to first connection path and the second connection path.
Load-balancing device can make particularly for first device to be connected to numerous autonomous systems of data communication network
Obtain the different connection paths that can pass through the offer of different autonomous systems from first device to second device.
Hereinafter, exemplary embodiment of the present invention is described with reference to schema.
Detailed description of the invention
Fig. 1 schematically shows the step of method of an exemplary embodiment of the present invention.
Fig. 2 schematically shows the step of method of an exemplary embodiment of the present invention.
Fig. 3 schematically shows the data communication network for implementing the method for an exemplary embodiment of the present invention.
Fig. 4 schematically shows the data communication network for implementing the method for an exemplary embodiment of the present invention.
Fig. 5 schematically shows the load-balancing device of an exemplary embodiment of the present invention.
Fig. 6 shows the end to end connection between first device and second device.
Specific embodiment
Schema is schematical and is not necessarily drawn to scale.Similar reference symbol is related to similar or identical element.
The step of Fig. 1 diagram is for being assigned to the method for numerous connection paths for load in data communication network.This side
Method is particularly suitable for including the data communication network for making at least two connection paths of numerous device interconnectings.Below with reference to Fig. 3 and 4
It shows and illustrates data communication network.
The method includes monitoring the data flow on the first connection path of data communication network and on the second connection path
Step S11.In subsequent step S12, the number in first connection path and the second connection path transport layer is determined
According to transport behavior.In the subsequent step S13, the first data throughout and the second connection path of first connection path are determined
Second data throughout.In subsequent step S14, load is distributed based on the first data throughout and the second data throughout
To first connection path and the second connection path.
Fig. 2 schematically shows the embodiment of some steps of the method according to Fig. 1.
Firstly, configure in the step s 21 such as Internetwork link data communication network link and corresponding parameter.Then
In step S22, all connection requests are tracked and according to such as destination and/or application type and/or priority by the company
Requests classification is connect into each group.In step S23, each group is distributed in all available connections of data communication network, that is,
Each available connection is assigned to by each group.In order to determine the link-quality of connection path, it is directed to each link in step s 24
On each group of measurement Connecting quantity and establish baseline for each group of each chain road.
Now, in step s 25, determine whether any group of the parameter on specific chain road shows exception.If do not deposited
In exception, then repeatedly abovementioned steps S22 to S24.If there is exception, then being determined on the same link in step S26
Whether other groups show identical or other exceptions.
If other groups show exception, may infer that in step s 27 in the presence of such as ISP
The connection error of (Internet Service Provider, ISP) connection error, and use link protection.If in same chain
Other groups of road do not show exception, then determining that identical group of other connections on other chain roads are in step S28
The similar exception of no display.
If it is then may infer that destination server in step S29, that is, requested terminal installation, it may be not quasi-
It gets ready and service or may be disabled.In the case, continue to track.If its of identical group on other chain roads
It connects and does not show similar exception, then may infer that in step s 30 it is abnormal be thus connected path show and will described in
Group is switched to other links, that is, the connection path for a connection group is changed.
Fig. 3 schematically shows tool, and there are three the data communication networks of terminal installation T1, T2 and T3 (terminal 1 to 3).Terminal dress
It sets and is interconnected by the link of data communication network.In general, data communication network is made of node and edge, interior joint and pass through edge
Autonomous system AS1, AS2, AS3 and AS4 of interconnection are corresponding, and the edge is that is, the number for interconnecting terminal installation and autonomous system
According to link 130A, 130B, 130C, 130D.
As shown in Figure 3, different connection paths can be used for specific end to end connection.For example, being filled in terminal
Setting the end to end connection between T1 and T2 can be by establishing from T1 to AS1 to the first connection path 130A of T2.Alternatively, phase
Same end to end connection can be by establishing from T1 to AS2 to the second connection path 130B of T2.It such as can be clear from schema
See to Chu, connection path in addition be it is possible, as T1 to AS3 arrive AS2 to T2.
Also indicate between T1 and T3 by the third connection path 130C of AS2 and for from T1 to AS3 to AS4 to T3's
4th connection path 130D of this end to end connection.
Terminal installation T1, T2, T3 can be used for executing method as described in this article.Particularly, terminal installation T1,
T2, T3 may include load-balancing device, be adapted for carrying out corresponding method and step to determine to given terminal installation
Preferred connection path.
Method as described in this article and load equipment can be provided in the preferable connection between terminal installation, because
It determines when the data which connection path is used for another terminal installation are transmitted in view of the autonomous system of data communication network
Link-quality and load.
Fig. 4 is shown similar to the data communication network of the data communication network of Fig. 3.Autonomous system is indicated as passing through data link
Wide area network 150A, 150B, 150C, 150D interconnected amongst one another.It is multiple and different between client 110 and server 120A, 120B
Connection path is possible.
Client 110 is connected to data communication network for example, by the data network access node 140 of more wan routers etc..
More wan routers are connected to more than one WAN of data communication network, and therefore can will be from client 110 by WAN 150B
Data flow to first server 120A is conducted directly to first server 120A or is directed to WAN 150A by WAN 150D
To WAN 150B and then to first server 120A.If the data link between wan router 140 and WAN 150B has
There is low quality, then latter connection path can be preferably.
Similarly, two different connection paths of second server 120B can be used in client 110.One connection
Path can be established by WAN 150D and 150C.Alternatively, WAN 150D can pass through WAN 150B to WAN 150A to WAN
It 150C and is bypassed to second server 120B.
More wan routers 140 may include load-balancing device as described with reference to Figure 5.
Fig. 5 shows load-balancing device 200.Load-balancing device 200 includes connecing for being connected to the first of terminal installation
The second interface 204 of 202 and numerous WAN for being connected to data communication network of mouth.Load-balancing device further comprises connecting
Connect tracking module 210, packet inspection module 220, storage unit 230, analytical unit 240 and Load Balance Unit 250.
Connection tracking module 210 is used to monitor the data flow on first connection path and on the second connection path, example
Such as, as shown in Figure 4 from client 110 to the data flow of first server 120A.Analytical unit 240 is for determining first
Data transport behavior in connection path and the second connection path transport layer.Connection path is usually assigned to data communication
The different WAN of net, that is, be assigned to the different output port of second interface 204.Analytical unit 240 is further used for determining first
First data throughout of connection path and the second data throughout of the second connection path.Load Balance Unit 250 is used for base
Load is assigned to first connection path and the second connection path in the first data throughout and the second data throughout.In other words
It says, the output port of second interface 204 is assigned to data flow by Load Balance Unit.
Load-balancing device 200 further comprises packet inspection module 220 and storage unit 230, for example, database.
Packet inspection module 220 is for realizing the fine grit classification for outputting and inputting data packet and data flow.Storage unit 230 is used for
Store the strategy and information about data flow, connection path and its statistical data or parameter.
Fig. 6 shows the end to end connection between terminal installation T1 and T2.These terminal installations pass through data link 130
(such as 130A, 130B) physically interconnects (wired combination or wireless connection).Terminal installation respectively includes communication stack
112,122.Communication stack includes the numerous functions of implementing in layer.Transport layer 114,124 for establishing in terminal installation T1 and
End to end connection between T2, and it is unrelated with the implementation on the lower layer of communication stack or upper layer.Transport layer is particularly by data communication network
Logic end to end connection 135 is established, the data communication network includes multiple nodes and edge, that is, physically separate link,
It shows as in Figures 3 and 4.
However, the method and load-balancing device of an exemplary embodiment of the present invention are particularly for assessing at two
The quality of end to end connection between (and only two) device, and it is unrelated with the physical implementation of lower data communication network.
List of numerals
The first autonomous system of AS1
T1 first device
110 clients
112 communication stacks
114 transport layers
T2 second device
120A first server
120B second server
122 communication stacks
124 transport layers
130 data link
130A first connection path
The second connection path of 130B
135 logical links
140 data network access nodes (more wan routers)
150 autonomous systems
200 load-balancing devices
202 first interfaces
204 second interfaces
210 connection tracking modules
220 packet inspection modules
230 storage units
240 analytical units
250 Load Balance Unit
Claims (26)
1. a kind of for load to be assigned to numerous connection path (130A, 130B, 130C, 130D) in data communication network
Method,
It is characterized in that, the terminal installation that data communication network allows attachment to data communication network passes through the link road of data communication network
Diameter exchanges data；Data communication network includes multiple wide area networks, and the wide area network is part, region or logic of data communication network
System；
The data communication network includes at least two connection paths, that is, interconnects first device (T1) and second device (T2)
First connection path (130A) and the second connection path (130B)；
It the described method comprises the following steps:
Monitor the data flow of (S11) on the first connection path (130A) and on second connection path (130B)；
Determine (S12) the first connection path (130A) and described second connection path (130B) transport layer (114,
124) the data transport behavior on；
Determine first data throughout and second connection path (130B) of (S13) described first connection path (130A)
Second data throughout；
First link road is arrived into load distribution (S14) based on first data throughout and second data throughout
Diameter (130A) and second connection path (130B)；
Wherein, biography of the determination (S12) in the first connection path (130A) and described second connection path (130B)
The step of data transport behavior on defeated layer (114,124), comprises determining that the first connection path and second link road
The TCP session behavior of diameter；
Wherein, the step of TCP session behavior of the determination first connection path and second connection path includes:
Determine the window size behavior of the TCP session at the appointed time section；It determines and is received by the first device (T1)
The first data frame and the second data frame between time delay；Wherein first data frame and second data frame are described
The continuous data frame of TCP session.
2. according to the method described in claim 1,
It is characterized in that, the monitoring (S11) is on the first connection path (130A) and in second connection path
The step of data flow on (130B) includes:
Monitor the input traffic of the first device (T1).
3. according to the method described in claim 1,
It is characterized in that, the monitoring (S11) is on the first connection path (130A) and in second connection path
The step of data flow on (130B) includes:
Monitor the output stream of the first device (T1).
4. according to the method described in claim 2,
It is characterized in that, the monitoring (S11) is on the first connection path (130A) and in second connection path
The step of data flow on (130B) includes:
Monitor the output stream of the first device (T1).
5. according to claim 1 to method described in any one of 4,
It is characterized in that, the data communication network includes that the third for interconnecting the first device (T1) and 3rd device (T3) connects
Connect path (130C) and the 4th connection path (130D)；
Wherein in any one of above claim with the first connection path (130A) and second connection path
The step of (130B) correlation is separately to the third connection path (130C) and the 4th connection path (130D).
6. according to the method described in claim 5,
It is characterized in that, further including steps of
The numerous connection paths for coming from the first device are grouped by the device depending on being connected, so that will be from described first
Numerous connection paths of device to the second device are grouped into first group and will fill from the first device to the third
The numerous output connection paths set are grouped into second group；
The step related to the first connection path (130A) and the second connection path (130B) is applied respectively
To described first group and second group of the connection path.
7. according to claim 1 to method described in any one of 4,
It is characterized in that, the step of TCP session behavior of the determination first connection path and second connection path
Include:
It determines in the request for being sent to the second device (T2) by the first device (T1) and comes from the second device
(T2) the time delay between response.
8. according to the method described in claim 7,
It is characterized in that, the step of TCP session behavior of the determination first connection path and second connection path
Include:
Determine the sequence number by the received continuous data frame of the first device (T1)；And
Identify whether to lose a certain frame in connection path.
9. according to the method described in claim 5,
It is characterized in that, the step of TCP session behavior of the determination first connection path and second connection path
Include:
It determines in the request for being sent to the second device (T2) by the first device (T1) and comes from the second device
(T2) the time delay between response.
10. according to the method described in claim 6,
It is characterized in that, the step of TCP session behavior of the determination first connection path and second connection path
Include:
Determine the sequence number by the received continuous data frame of the first device (T1)；And
Identify whether to lose a certain frame in connection path.
11. according to claim 1 to method described in any one of 4,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
12. according to the method for claim 11,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
13. according to the method described in claim 5,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
14. according to the method for claim 13,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
15. according to the method described in claim 6,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
16. according to the method for claim 15,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
17. according to the method described in claim 7,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
18. according to the method for claim 17,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
19. according to the method described in claim 8,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
20. according to the method for claim 19,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
21. according to the method described in claim 9,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
22. according to the method for claim 21,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
23. according to the method described in claim 10,
It is characterized in that, further including steps of
Machine learning algorithm is applied to being supervised for the first connection path (130A) and the second connection path (130B)
Depending on data flow, to find that the connection on the first connection path (130A) and second connection path (130B) is asked
Topic.
24. according to the method for claim 23,
It is characterized in that, described arrive the first connection path (130A) and second connection path for load distribution (S14)
(130B) is executed based on the machine learning algorithm.
25. a kind of for load to be assigned to numerous connection paths (130A, 130B, 130C, 130D) in data communication network
Load-balancing device (200),
It is characterized in that, the terminal installation that data communication network allows attachment to data communication network passes through the link road of data communication network
Diameter exchanges data；Data communication network includes multiple wide area networks, and the wide area network is part, region or logic of data communication network
System；
The data communication network includes at least two connection paths, that is, interconnects first device (T1) and second device (T2)
First connection path (130A) and the second connection path (130B)；
Wherein the load-balancing device includes:
It connects tracking module (210), for monitoring (S11) on the first connection path (130A) and in second connection
Data flow on path (130B)；
Analytical unit (240), for determining (S12) in the first connection path (130A) and described second connection path
Data transport behavior in the transport layer (114,124) of (130B)；
Wherein the analytical unit (240) is further used for determining first data of (S13) described first connection path (130A)
Second data throughout of handling capacity and second connection path (130B)；And
Load Balance Unit (250), for being divided load based on first data throughout and second data throughout
The first connection path (130A) and second connection path (130B) are arrived with (S14)；
Wherein, biography of the determination (S12) in the first connection path (130A) and described second connection path (130B)
The step of data transport behavior on defeated layer (114,124), comprises determining that the first connection path and second link road
The TCP session behavior of diameter；
Wherein, the step of TCP session behavior of the determination first connection path and second connection path includes:
Determine the window size behavior of the TCP session at the appointed time section；It determines and is received by the first device (T1)
The first data frame and the second data frame between time delay；Wherein first data frame and second data frame are described
The continuous data frame of TCP session.
26. load-balancing device (200) according to claim 25 characterized by comprising
First interface (202), for being connected to first device (T1)；
Second interface (204), for being connected to data communication network；
Wherein the Load Balance Unit (250) is used to that load distribution (S14) to be arrived described the by the second interface (204)
One connection path (130A) and second connection path (130B).
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/EP2014/076636 WO2016087001A1 (en) | 2014-12-04 | 2014-12-04 | Load allocation in a data communication network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105900385A CN105900385A (en) | 2016-08-24 |
| CN105900385B true CN105900385B (en) | 2019-08-27 |
Family
ID=52011207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201480036848.8A Active CN105900385B (en) | 2014-12-04 | 2014-12-04 | Load distribution in data communication network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105900385B (en) |
| WO (1) | WO2016087001A1 (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1756233A (en) * | 2004-09-30 | 2006-04-05 | 富士通株式会社 | Route selection method in the communication network and device |
| CN101754319A (en) * | 2008-12-10 | 2010-06-23 | 华为技术有限公司 | Routing method, routing device and routing system |
| US7903557B1 (en) * | 2005-09-21 | 2011-03-08 | Qlogic, Corporation | Multipathing using multiple endpoint addresses for load balancing in a network |
| CN103312612A (en) * | 2012-03-06 | 2013-09-18 | 张昱 | Optimizing method and system for multi-constrain quality-of-service routing |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NO873860L (en) * | 1986-10-31 | 1988-05-02 | Asta Pharma Ag | IFOSPHAMIDE LYOPHILIZATE AND PROCEDURE FOR THEIR PREPARATION. |
| US9071541B2 (en) * | 2012-04-25 | 2015-06-30 | Juniper Networks, Inc. | Path weighted equal-cost multipath |
-
2014
- 2014-12-04 WO PCT/EP2014/076636 patent/WO2016087001A1/en active Application Filing
- 2014-12-04 CN CN201480036848.8A patent/CN105900385B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1756233A (en) * | 2004-09-30 | 2006-04-05 | 富士通株式会社 | Route selection method in the communication network and device |
| US7903557B1 (en) * | 2005-09-21 | 2011-03-08 | Qlogic, Corporation | Multipathing using multiple endpoint addresses for load balancing in a network |
| CN101754319A (en) * | 2008-12-10 | 2010-06-23 | 华为技术有限公司 | Routing method, routing device and routing system |
| CN103312612A (en) * | 2012-03-06 | 2013-09-18 | 张昱 | Optimizing method and system for multi-constrain quality-of-service routing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105900385A (en) | 2016-08-24 |
| WO2016087001A1 (en) | 2016-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gilly et al. | An up-to-date survey in web load balancing | |
| CN104272708B (en) | It is distributed with the stateless first order grouping to server farm and is distributed to the secondary data packets of the stateful second level grouping distribution of some server in group | |
| CN102624584B (en) | Chain circuit detecting method and device | |
| US8477772B2 (en) | System and method for determination of routing information in a network | |
| US11044199B2 (en) | Inferring device load and availability in a network by observing weak signal network based metrics | |
| CN106656847A (en) | Software defined network (SDN) load balancing method with highest network utility | |
| CN113315682A (en) | Method, system and apparatus for generating information transmission performance warning | |
| CN104363159A (en) | Virtual open network building system and method based on software definition network | |
| US10833974B2 (en) | Dynamic configuration of routing paths in a data network | |
| JP2009543500A (en) | Connectivity fault management (CFM) in networks with link aggregation group connections | |
| CN105471609A (en) | Method and device for configuring business | |
| CN103259809A (en) | Load balancer, load balancing method and stratified data center system | |
| CN104270291A (en) | Content delivery network (CDN) quality monitoring method | |
| EP2876844A1 (en) | Self-debugging router platform | |
| EP2767038A1 (en) | Detection of load balancing across network paths in a communication network | |
| EP2586158B1 (en) | Apparatus and method for monitoring of connectivity services | |
| US7907599B2 (en) | Determination of SIP transport to reduce call setup delays | |
| US10623278B2 (en) | Reactive mechanism for in-situ operation, administration, and maintenance traffic | |
| CN103891217A (en) | Service assurance using network measurement triggers | |
| CN103262046A (en) | Server management apparatus, server management method, and program | |
| CN102823194A (en) | Operations, administrations and management proxy and a method for handling operations, administrations and management messages | |
| CN114513467B (en) | Network traffic load balancing method and device of data center | |
| JP4678652B2 (en) | P2P traffic monitoring control apparatus and method | |
| CN108353027A (en) | A kind of software defined network system for detecting port failure | |
| CN108494625A (en) | A kind of analysis system on network performance evaluation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220207 Address after: 550025 Huawei cloud data center, jiaoxinggong Road, Qianzhong Avenue, Gui'an New District, Guiyang City, Guizhou Province Patentee after: Huawei Cloud Computing Technology Co.,Ltd. Address before: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen Patentee before: HUAWEI TECHNOLOGIES Co.,Ltd. |
|
| TR01 | Transfer of patent right |