CN106357726B - Load-balancing method and device - Google Patents
Load-balancing method and device Download PDFInfo
- Publication number
- CN106357726B CN106357726B CN201610718285.1A CN201610718285A CN106357726B CN 106357726 B CN106357726 B CN 106357726B CN 201610718285 A CN201610718285 A CN 201610718285A CN 106357726 B CN106357726 B CN 106357726B
- Authority
- CN
- China
- Prior art keywords
- session
- data processing
- processing unit
- information
- data
- 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
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/14—Session management
- H04L67/148—Migration or transfer of sessions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Computer And Data Communications (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a kind of load-balancing method and devices, are related to field of communication technology, and the load balancing in data processor between data processing unit may be implemented.The described method includes: the session for obtaining data processing unit in data processor distributes information；Information and default load balance scheduling rule are distributed according to the session, generates the conversation dispatching information of load balancing between data processing unit；The conversation shift in the data processor between different data processing units is carried out according to the conversation dispatching information.The present invention is suitable for load balance process.
Description
Technical field
The present invention relates to a kind of fields of communication technology, more particularly to a kind of load-balancing method and device.
Background technique
With the continuous development of the communication technology, network interface card technology is more and more mature.Network interface card is group of networks of the work in link layer
Part is the interface that computer and transmission medium are connected in local area network, the physics being not only able to achieve between local network transport medium
Connection and electric signal matching, further relate to the sending and receiving of frame, the encapsulation of frame and opening, medium access control, data coding
With decoding and the function of data buffer storage etc..
Currently, a new generation network interface card usually support it is multiple send and receive queue, i.e., more queue network interface cards.The network interface card can incite somebody to action
Received network packet is distributed to different queues, so that network flow can be in a data processor at different data
Manage parallel processing between unit.Specifically, for network interface card by packet classification algorithm, which point be Lai Zhiding network packet belong to
Class, the data packet of each classification are sent in a specified receiving queue, and data processor can be to the data in each queue
Packet is handled, the data packet in one queue of each data processing unit alignment processing, so that each data processing list
Member handles respective session task.
However, it is non-uniform to have data stream distribution when carrying out packet delivery by above-mentioned packet classification algorithm
Situation, for example, the input parameter as hash function carries out Hash operation, and according to meter according to data packet respective field is extracted
The function result of calculating is distributed data packet, and session task can be distributed to different data processing units at random, will lead to number
According to existing simultaneously the excessive data processing unit of session task and the very few data processing units of session task in processor.In turn
Will cause the unbalanced problem of the number of sessions handled between data processing unit, so as to cause different data processing units it
Between load imbalance, affect the treatment effeciency of data processor.
Summary of the invention
In view of this, main purpose is can be real the embodiment of the invention provides a kind of load-balancing method and device
Load balancing between existing different data processing units, can be improved the treatment effeciency of data processor.
According to an embodiment of the present invention on one side, a kind of load-balancing method is provided, this method comprises:
The session for obtaining data processing unit in data processor distributes information；
Information and default load balance scheduling rule are distributed according to the session, it is equal to generate load between data processing unit
The conversation dispatching information of weighing apparatus；
It is moved according to the session that the conversation dispatching information carries out in the data processor between different data processing units
It moves.
Further, the number comprising data processing unit in the data processor in the session distribution information, with
And the currently processed number of sessions of each data processing unit in the data processor, it is described to distribute information according to the session
With default load balance scheduling rule, the conversation dispatching information of load balancing between data processing unit is generated, is specifically included:
According to the number of sessions, the number of data processing unit and default load balance scheduling rule, every number is determined
According to the session for needing to migrate in processing unit and former data processing unit corresponding with the session for needing to migrate and target data
Processing unit；
The session migrated and former data processing unit corresponding with the session for needing to migrate and target are needed according to described
Data processing unit generates the conversation dispatching information of load balancing between data processing unit.
Further, described that different data processing units in the data processor are carried out according to the conversation dispatching information
Between conversation shift, specifically include:
In the data accepted packet after treatment for the session that former data processing unit migrates needs, according to the session
Scheduling information, by the conversation shift for needing to migrate to target data processing unit.
Further, described that different data processing units in the data processor are carried out according to the conversation dispatching information
Between conversation shift after, the method also includes:
Record the identification information of the corresponding target data processing unit of session of migration；
When receiving data packet corresponding with the session of the migration, according to the identification information, by the data packet
It is assigned in the target data processing unit and is handled.
Further, described if the currently employed session Hash table of the data processor is local session Hash table
Method further include:
The session information data field that will be migrated session is copied to be migrated at the corresponding target data of session with described
In the session Hash table for managing unit.
Further, the method also includes:
After the completion of conversation shift or when establishing new Dialog processing task, to data processing list in the data processor
The session distribution information of member is updated.
Further, the session for obtaining data processing unit in data processor distributes information, specifically includes:
Information is distributed according to the session that prefixed time interval obtains data processing unit in the data processor.
According to an embodiment of the present invention on the other hand, a kind of load balancing apparatus is provided, which includes:
Acquiring unit, the session for obtaining data processing unit in data processor distribute information；
Generation unit, session distribution information and default load balance scheduling rule for being obtained according to the acquiring unit
Then, the conversation dispatching information of load balancing between data processing unit is generated；
Migration units, the conversation dispatching information for being generated according to the generation unit carry out in the data processor not
With the conversation shift between data processing unit.
Further, the number comprising data processing unit in the data processor in the session distribution information, with
And the currently processed number of sessions of each data processing unit in the data processor, the generation unit specifically include:
Determining module, for according to the number of the number of sessions, data processing unit and default load balance scheduling rule
Then, the session for needing to migrate in each data processing unit and former data processing list corresponding with the session for needing to migrate are determined
Member and target data processing unit；
Generation module, the session that the needs for being determined according to the determining module migrate and the session migrated with needs
Corresponding original data processing unit and target data processing unit, generate the conversation dispatching of load balancing between data processing unit
Information.
The migration units, specifically at the data accepted packet for the session that former data processing unit migrates needs
After reason, according to the conversation dispatching information, by the conversation shift for needing to migrate to target data processing unit.
Further, described device further include:
Recording unit, the identification information of the corresponding target data processing unit of session for recording migration；
Allocation unit, for when receiving data packet corresponding with the session of the migration, according to the recording unit
The identification information of record handles the allocation of packets into the target data processing unit.
Further, described device further include:
Copied cells, if the session Hash table currently employed for the data processor is local session Hash table,
The session information data field that will be migrated session copies to and the session corresponding target data processing unit of being migrated
In session Hash table.
Further, described device further include:
Updating unit, for after the completion of conversation shift or when establishing new Dialog processing task, to the data processing
The session distribution information of data processing unit is updated in device.
The acquiring unit is specifically used for obtaining data processing unit in the data processor according to prefixed time interval
Session distribute information.
By above-mentioned technical proposal, technical solution provided in an embodiment of the present invention is at least had the advantage that
A kind of load-balancing method and device provided in an embodiment of the present invention, first data processing in acquisition data processor
The session of unit distributes information；Then information and default load balance scheduling rule are distributed according to the session, generated at data
Manage the conversation dispatching information of load balancing between unit；Finally carried out in the data processor according to the conversation dispatching information
Conversation shift between different data processing units.Session of the embodiment of the present invention according to data processing unit in data processor
Information and default load balance scheduling rule are distributed, generates conversation dispatching information, and session is executed according to the conversation dispatching information
Migration operation, the number of sessions to realize each data processing unit in data processor is balanced, and then difference may be implemented
Load balancing between data processing unit, so as to improve the treatment effeciency of data processor.
The above description is only an overview of the technical scheme of the present invention, in order to better understand the technical means of the present invention,
And it can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage can
It is clearer and more comprehensible, the followings are specific embodiments of the present invention.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention
Limitation.And throughout the drawings, the same reference numbers will be used to refer to the same parts.In the accompanying drawings:
Fig. 1 shows a kind of load-balancing method flow diagram provided in an embodiment of the present invention；
Fig. 2 shows another load-balancing method flow diagrams provided in an embodiment of the present invention；
Fig. 3 shows a kind of specific processing flow schematic diagram provided in an embodiment of the present invention；
Fig. 4 shows a kind of load balancing apparatus structural schematic diagram provided in an embodiment of the present invention；
Fig. 5 shows another load balancing apparatus structural schematic diagram provided in an embodiment of the present invention；
Fig. 6 shows a kind of schematic diagram of SiteServer LBS framework provided in an embodiment of the present invention.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
The embodiment of the invention provides a kind of load-balancing method, the load between different data processing units may be implemented
Equilibrium, as shown in Figure 1, which comprises
101, the session for obtaining data processing unit in data processor distributes information.
Wherein, the processing for network packet main security business, safety service may be implemented in the data processor
Not only have a basic connectivity energy function, such as exchange, routing, agency, can also include function of safety protection, as access control,
Attack defending, anti-virus, application control etc..The data processor may include multiple data processing units, for example, X86 frame
Multi-core CPU (Central Processing Unit, central processing unit), ASIC (the Application Specific of structure
Integrated Circuit, integrated circuit) framework packet processing chip group etc., each data processing unit is according to business demand
Network data stream, can specifically execute and establish the operations such as session, forwarding data packet, and each data processing unit storage is respective
Session information data.
The session, which distributes in information, may include total sessions currently processed in data processor, in data processor
The currently processed number of sessions etc. of each data processing unit in the number and data processor of data processing unit.
In embodiments of the present invention, the meeting of data processing unit in data processor can be obtained from monitoring log information
Words distribute information, have recorded the currently processed session task distribution condition of data processor in the monitoring log information.For example, from
Conversation analysis information is acquired in monitoring log information, which distributes in information comprising in data processor currently processed 10
A session task, the data processor have 3 data processing units, respectively No. 1 unit, No. 2 units, No. 3 units, No. 1 list
The currently processed 4 session tasks of member, the currently processed 1 Dialog processing task of No. 2 units, currently processed 5 sessions of No. 3 units are appointed
Business.
102, information and default load balance scheduling rule are distributed according to session, it is equal generates load between data processing unit
The conversation dispatching information of weighing apparatus.
Wherein, may include in the conversation dispatching information need migrate session, with the needs migrate session it is corresponding
Former data processing unit and the information such as target data processing unit.For example, the form that can use data mapping tables saves life
At conversation dispatching information, every a line in the data mapping tables includes a session for needing to migrate, and with this session
The corresponding original fields such as data processing unit and target data processing unit.
The default load balance scheduling rule can be carried out writing configuration by technical staff in advance.In the embodiment of the present invention
In, based on default load balance scheduling rule, using current session distribution condition as parameter, corresponding scheduling can be executed and calculated
Method generates the conversation dispatching information of load balancing between data processing unit.It should be noted that for the embodiment of the present invention,
It can support a variety of dispatching algorithms, and since conversation shift process expense is larger, dispatching algorithm needs to consider migration overhead, with
Just guarantee whole conversation shift efficiency.
For example, 10 session tasks currently processed comprising data processor in the session distribution information acquired, it should
Data processor has 3 data processing units, respectively No. 1 unit, No. 2 units, No. 3 units, and No. 1 unit currently processed 6
Session task, the currently processed 3 Dialog processing tasks of No. 2 units, the currently processed 1 session task of No. 3 units, at this time 3 numbers
It is unbalanced according to the number of sessions handled between processing unit, it will lead to load imbalance between 3 data processing units.To understand
The certainly above problem, using current sessions distribution condition as parameter, executes load balancing according to default load balance scheduling rule
Dispatching algorithm calculates the session for needing to migrate and target data processing unit etc., and generates conversation dispatching information, the session tune
Include two session tasks for needing to migrate in No. 1 unit, and former data corresponding with the two session tasks in degree information
Processing unit (i.e. No. 1 unit) and target data processing unit (i.e. No. 3 units), to realize two meetings in No. 1 unit
Words task immigration is handled into No. 3 units, and then realizes the load balancing between 3 data processing units.
It in embodiments of the present invention, can also include: to distribute information determination according to session to work as before executing step 102
The least data processing unit of pre-treatment session task amount is executed the behaviour for generating conversation dispatching information by the data processing unit
Make, avoids executing the operation for generating conversation dispatching information in the more data processing unit of currently processed session task amount, into
And the load pressure of data processor can be mitigated.
103, the conversation shift in data processor between different data processing units is carried out according to conversation dispatching information.
For the embodiment of the present invention, during conversation shift, session information data can be migrated, wherein
It may include five-tuple information, data packet count information, disengaging port information, session state information in the session information data
Deng.
In embodiments of the present invention, according to conversation dispatching information, the session of data processing unit in data processor is manipulated
Task, it will words task moves to another data processing unit from a data processing unit.
A kind of load-balancing method provided in an embodiment of the present invention, data processing unit first in acquisition data processor
Session distributes information；Then information and default load balance scheduling rule are distributed according to the session, generates data processing unit
Between load balancing conversation dispatching information；Different numbers in the data processor are finally carried out according to the conversation dispatching information
According to the conversation shift between processing unit.The embodiment of the present invention distributes letter according to the session of data processing unit in data processor
Breath and default load balance scheduling rule generate conversation dispatching information, and execute conversation shift behaviour according to the conversation dispatching information
Make, the number of sessions to realize each data processing unit in data processor is balanced, and then may be implemented at different data
Load balancing between unit is managed, so as to improve the treatment effeciency of data processor.
In order to preferably understand above-mentioned method shown in FIG. 1, as the refinement and extension to above embodiment,
The embodiment of the invention provides another load-balancing methods, as shown in Figure 2, which comprises
201, the session for obtaining data processing unit in data processor distributes information.
After more queue network interface cards receive data packet from network, all network packets are classified, different classes of data
Stream is sent to the different data processing units in data processor.In order to realize load balancing between different data processing units,
The session of data processing unit distributes information in available data processor, and conversates and move according to session distribution information
It moves.
202, according to the currently processed number of sessions of each data processing unit, the number of data processing unit and preset negative
Balance dispatching rule is carried, determines the session for needing to migrate in each data processing unit and corresponding with the session that migrates is needed
Former data processing unit and target data processing unit.
For example, 10 session tasks currently processed comprising data processor in the session distribution information acquired, it should
Data processor has 3 data processing units, respectively No. 1 unit, No. 2 units, No. 3 units, and No. 1 unit currently processed 5
Session task, the currently processed 0 session task of No. 2 units, the currently processed 5 session tasks of No. 3 units, at this time at 3 data
The number of sessions handled between reason unit is unbalanced, will lead to load imbalance between 3 data processing units.On solving
Problem is stated, according to default load balance scheduling rule, with the currently processed number of sessions of each data processing unit, data processing
The sessions such as number of unit distribution condition executes the dispatching algorithm of load balancing as parameter, calculate need migrate session,
Former data processing unit, target data processing unit.Specifically, it is determined that a session task for needing to migrate in No. 1 unit, and
Original data processing unit (i.e. No. 1 unit) corresponding with this session task and target data processing unit (i.e. No. 2 units)；Really
Two session tasks for needing to migrate in fixed No. 3 units, and former data processing unit (i.e. 3 corresponding with the two session tasks
Number unit) and target data processing unit (i.e. No. 2 units).
203, the session and former data processing unit corresponding with the session for needing to migrate that migrate as needed and target
Data processing unit generates the conversation dispatching information of load balancing between data processing unit.
For example, 16 session tasks that data processor is currently processed, the data processor have 4 data processing units,
Respectively No. 1 unit, No. 2 units, No. 3 units, No. 4 units, the currently processed 7 session tasks of No. 1 unit, No. 2 units are current
Handle 4 session tasks, the currently processed 5 session tasks of No. 3 units, the currently processed 0 session task of No. 4 units, for reality
The number of sessions handled between this existing 4 data processing units is balanced, according to default load balance scheduling rule, with current sessions
Distribution condition executes the dispatching algorithm of load balancing as parameter, calculates the session for needing to migrate and target data processing is single
Member etc., and conversation dispatching information is generated, it include three session tasks for needing to migrate in No. 1 unit in the conversation dispatching information,
And former data processing unit (i.e. No. 1 unit) corresponding with these three session tasks and target data processing unit (i.e. No. 4 lists
Member), also comprising a session task for needing to migrate in No. 3 units, and former data processing list corresponding with this session task
First (i.e. No. 3 units) and target data processing unit (i.e. No. 4 units).
204, the conversation shift in data processor between different data processing units is carried out according to conversation dispatching information.
Further, in order to which the number of sessions for realizing each data processing unit in data processor is balanced and then right
The distribution processor of data flow reaches relatively uniform state, gives full play to the processing energy of all data processing units of data processor
Power maximizes the treatment effeciency of data processor.
During conversation shift, session information saves complete, the process phase with process migration, virtual machine (vm) migration etc.
Than transition process is simple, and transport efficiency is high, and there is no data processing units to handle stationary problem.For example, conversation shift
Process is the operation in data processor between data processing unit, be not data processing unit jointly from memory etc. otherly
Side obtains data and is handled, and there is no data processing units to handle stationary problem.
Specifically, the step 204 can specifically include: in having connect for the session that former data processing unit migrates needs
Data packet after treatment is received, according to the conversation dispatching information, needs the conversation shift migrated at target data for described
Manage unit.In conversation procedure, session task is usually made of continuous data packet stream, these data packets can according to generation when
Sequence is distributed serially to be handled to the data processing unit of the corresponding session task.In the present embodiment, when the migration that conversates
When, data packet is had been received for former data processing unit, still by former data processing unit before conversation shift to its into
Row processing, after by session task immigration to target data processing unit, the subsequent new data packet generated again of the session task
Then by target data processing unit processes.In the prior art, after session task is assigned to some data processing unit, the meeting
All data packets of words task are handled by the corresponding data processing unit.The present embodiment can be in processing session data
Session task is migrated during packet, and transition process not will cause the interruption of session traffic.
It should be noted that in data accepted packet of the former data processing unit to the session for needing to migrate, processing terminate
Afterwards, then the migration operation of the session is carried out, this migration operation may be implemented under the premise of not interrupting session traffic processing, will
Session moves to another data processing unit from a data processing unit, during conversation shift, corresponding session
Data packet is not in stream packet, postpones the problems such as excessive, synchronous, so that forwarding data flow is unaffected in transition process.
In embodiments of the present invention, it can be carried out according to the classification information for the session Hash table that data processor uses specific
Conversation shift operation, wherein the classification information can be global session Hash table classification, local session Hash table classification.
If the currently employed session Hash table of data processor is global session Hash table, illustrate in the data processor at each data
The session Hash table content for managing unit use is identical, and the session information data field for being migrated session is not necessarily to from former data
It manages and is copied in the session Hash table of target data processing unit in the session Hash table of unit, because there are identical session letters
Data field is ceased, therefore, directly conversate migration step.
If the currently employed session Hash table of data processor is local session Hash table, illustrate every in the data processor
The session Hash table content that a data processing unit uses is not identical.Further, in order to guarantee target data processing unit
There is the session information data field for being migrated session in session Hash table, the method can also include: that will be migrated session
Session information data field copy to in the session Hash table for being migrated the corresponding target data processing unit of session.
So that target data processing unit is according to the session information data field of the duplication, continue with that be migrated session corresponding with this
Data packet.It should be noted that the mode of copy session information data field is without being limited thereto, the meeting of session can also be will be migrated
Words information data field, which is clipped to, to be migrated in the session Hash table of the corresponding target data processing unit of session with this.
Further, after in order to guarantee conversation shift, data processor receives network packet again and is able to carry out accordingly
Processing, after step 204, the method can also include: the corresponding target data processing unit of session of record migration
Identification information, wherein the identification information can be title, the ID (Identity, identity number) of data processing unit
Number etc.；When receiving data packet corresponding with the session of the migration, according to the identification information, by the allocation of packets
It is handled into the target data processing unit.
Specifically, described according to the identification information, by the allocation of packets into the target data processing unit
The step of being handled can specifically include: pass through preset distributor or former data processing list corresponding with the session of the migration
Member handles the allocation of packets into the target data processing unit.Wherein, the preset distributor can be
With the device that the session data newly received is forwarded a packet to corresponding data processing unit function.
In embodiments of the present invention, when receiving data packet corresponding with the session of migration, if there is no above-mentioned pre-
Distributor is set, it can also be according to the identification information of the target data processing unit of record, by former data corresponding with the session
Unit forwards are managed to being handled in target data processing unit, will not influence the processing of session task.
Further, in order to guarantee the real-time of load balancing between data processing unit in data processor, the side
When method can also include: after the completion of conversation shift or establish new Dialog processing task, to data in the data processor
The session distribution information of processing unit is updated.
The newest session of data processing unit in the data processor, which is obtained, according to prefixed time interval distributes information.Its
In, the prefixed time interval can be configured in actual demand according to, for example, prefixed time interval can be set as 1 second,
2 seconds etc..It should be noted that by obtaining the newest session of data processing unit in data processor according to prefixed time interval
Information is distributed, may be implemented to obtain the newest session distribution information of data processing unit in data processor in real time, and according to this
Newest session distribution information executes conversation shift operation, so can with data processing unit in real-time ensuring data processor it
Between load balancing, maximize the treatment effeciency of data processor.
In embodiments of the present invention, for some special scenes, when such as carrying out properties of product assessment, it is desirable that assessment system is most
Bigization processing capacity can directly carry out packet delivery by network interface card, then execute the embodiment of the present invention in order to realize the purpose
The step 201 of offer is to process described in step 204, so that the session task quantity of each data processing unit is identical, Jin Erke
To carry out the assessment that system maximizes processing capacity under conditions of load balancing between data processing unit, set with passing through at present
The mode for setting the address pair specific network protocol (Internet Protocol, IP), which carries out data flow, uniformly distributes and compares, and is not required to
IP address pair is manually set, the balanced adjustment of the number of sessions between different data processing units can be realized automatically, improve
Convenience.
Process flow specific for the embodiment of the present invention can with as shown in figure 3,
Data packet receives: after network interface card receives data packet, according to the content information of data packet, data packet is pressed into flow point class, it will
Packet delivery is handled to different data processing units.
Data packet processing: each data processing unit can parse data packet, carry out corresponding every safety inspection,
And session information is saved, finally execute corresponding safe action.
Conversation dispatching: information is distributed according to the session of each data processing unit summarized, executes the scheduling of load balancing
Algorithm calculates the session for needing to migrate and former data processing unit corresponding with the session and target data processing unit.
Conversation shift: by the session of required migration, in the case where guaranteeing the unbroken situation of regular traffic, it will talk about from former number
According to processing unit dynamic migration to target data processing unit, entire transition process only influences two data processing units, to meeting
It is transparent for words.
Data packet is sent: data processing unit is after having carried out data packet processing, according to the information of the affiliated session of data packet,
It sends data packet in network by network interface card.
Another kind load-balancing method provided in an embodiment of the present invention, first data processing unit in acquisition data processor
Session distribute information；Then according to session distribute information and default load balance scheduling rule, generate data processing unit it
Between load balancing conversation dispatching information；Last foundation conversation dispatching information, in the meeting that former data processing unit migrates needs
The data accepted packet after treatment of words, then the migration operation of the session is carried out, it may be implemented do not interrupting at session traffic
Under the premise of reason, the load balancing between different data processing units, during conversation shift, the number of corresponding session are realized
The problems such as according to packet be not in packet loss, delay is excessive, synchronous, forwarding data flow is unaffected in transition process；It may be implemented
The newest session of data processing unit in data processor is obtained in real time and distributes information, and distributes information according to the newest session
Conversation shift operation is executed, and then can be maximum with the load balancing between data processing unit in real-time ensuring data processor
Change the treatment effeciency of data processor.
Further, the specific implementation as Fig. 1 the method, the embodiment of the invention provides a kind of load balancing dresses
It sets, as shown in figure 4, described device includes: acquiring unit 31, generation unit 32, migration units 33.
The acquiring unit 31 can be used for obtaining the session distribution information of data processing unit in data processor.Institute
The session that stating acquiring unit 31 can obtain in data processor from monitoring log information distributes information.The acquiring unit 31
It can be for session distributes the main functional modules of information in acquisition data processor in the present apparatus.Meeting is got in acquiring unit 31
After words distribution information, the generation unit 32 can be passed it to, and trigger the generation unit 32 and work.
The generation unit 32, the session distribution information and default load that can be used for being obtained according to the acquiring unit 31
Balance dispatching rule, generates the conversation dispatching information of load balancing between data processing unit.The generation unit 32 can be
Using current session distribution condition as parameter in the present apparatus, the dispatching algorithm of load balancing is executed, generates conversation dispatching information
Main functional modules.After generation unit 32 generates conversation dispatching information, the migration units 33 can be passed it to, and touch
It sends out migration units 33 described and carries out relevant work.
The migration units 33 can be used for carrying out the number according to the conversation dispatching information that the generation unit 32 generates
According to the conversation shift between different data processing units in processor.Further, each in data processor in order to realize
The number of sessions of data processing unit is balanced, and then reaches relatively uniform state to the distribution processor of data flow, gives full play to
The processing capacity of all data processing units of data processor, maximizes the treatment effeciency of data processor.The migration units
33 is conversate the main functional modules of migration operation in the present apparatus according to conversation dispatching information.Specifically, it will need to migrate
Session moved in target data processing unit from former data processing unit.
It should be noted that other of each functional unit involved by a kind of load balancing apparatus provided in an embodiment of the present invention
Corresponding description, can be with reference to the corresponding description in Fig. 1, and details are not described herein.
A kind of load balancing apparatus provided in an embodiment of the present invention, can be only fitted in terminal device, can be used for controlling
Conversation shift between data processing unit, specifically includes: acquiring unit, generation unit, migration units, single by obtaining first
The session that member obtains data processing unit in data processor distributes information；Then it is distributed by generation unit according to the session
Information and default load balance scheduling rule, generate the conversation dispatching information of load balancing between data processing unit；Finally lead to
It crosses migration units and carries out session in the data processor between different data processing units according to the conversation dispatching information
Migration.Session distribution information and default load balance scheduling of the embodiment of the present invention according to data processing unit in data processor
Rule generates conversation dispatching information, and executes conversation shift operation according to the conversation dispatching information, to realize data processor
In each data processing unit number of sessions it is balanced, and then load balancing between different data processing units may be implemented, from
And the treatment effeciency of data processor can be improved.
Further, the specific implementation as Fig. 5 the method, the embodiment of the invention provides another load balancing dresses
It sets, as shown in figure 5, described device includes: acquiring unit 41, generation unit 42, migration units 43.
The acquiring unit 41 can be used for obtaining the session distribution information of data processing unit in data processor.Institute
The session that stating acquiring unit 41 can obtain in data processor from monitoring log information distributes information.The acquiring unit 41
It can be for session distributes the main functional modules of information in acquisition data processor in the present apparatus.Meeting is got in acquiring unit 41
After words distribution information, the generation unit 42 can be passed it to, and trigger the generation unit 42 and work.
The generation unit 42, the session distribution information and default load that can be used for being obtained according to the acquiring unit 41
Balance dispatching rule, generates the conversation dispatching information of load balancing between data processing unit.The generation unit 42 can be
Using current session distribution condition as parameter in the present apparatus, the dispatching algorithm of load balancing is executed, generates conversation dispatching information
Main functional modules.After generation unit 42 generates conversation dispatching information, the migration units 43 can be passed it to, and touch
It sends out migration units 43 described and carries out relevant work.
The migration units 43 can be used for carrying out the number according to the conversation dispatching information that the generation unit 42 generates
According to the conversation shift between different data processing units in processor.The migration units 43 are in the present apparatus according to conversation dispatching
Information conversates the main functional modules of migration operation.Specifically, the session migrated will be needed to move from former data processing unit
It moves on in target data processing unit.
Optionally, it may include the number of data processing unit in the data processor in the session distribution information,
And the currently processed number of sessions of each data processing unit in the data processor.
Specifically, the generation unit 42 can specifically include: determining module 421, generation module 422.
The determining module 421 can be used for according to the number of sessions, the number of data processing unit and default load
Balance dispatching rule determines the session for needing to migrate in each data processing unit and original corresponding with the session for needing to migrate
Data processing unit and target data processing unit.
The generation module 422, the session that migrates of needs that can be used for being determined according to the determining module and with need
The corresponding former data processing unit of the session to be migrated and target data processing unit, it is equal to generate load between data processing unit
The conversation dispatching information of weighing apparatus.
The migration units 43, the reception number for the session that specifically can be used for migrating needs in former data processing unit
It is according to the conversation dispatching information, the conversation shift for needing to migrate is single to target data processing according to packet after treatment
Member.This migration operation may be implemented under the premise of not interrupting session traffic processing, it will talk about from a data processing unit
Move to another data processing unit.
Further, described device can also include: recording unit 44, allocation unit 45.
The recording unit 44 can be used for recording the mark letter of the corresponding target data processing unit of session of migration
Breath.
The allocation unit 45 can be used for when receiving data packet corresponding with the session of the migration, according to institute
The identification information for stating recording unit records handles the allocation of packets into the target data processing unit.Into
And after can guaranteeing conversation shift, data processor receives network packet again and is able to carry out respective handling.
Further, described device can also include: copied cells 46.
The copied cells 46, if can be used for the currently employed session Hash table of the data processor for local session
Hash table, then the session information data field that will be migrated session is copied to be migrated at the corresponding target data of session with described
In the session Hash table for managing unit.It can guarantee to exist in the session Hash table of target data processing unit in turn and be migrated session
Session information data field continued with so that target data processing unit is according to the session information data field of the duplication
The corresponding data packet of session is migrated with this.
Further, described device can also include: updating unit 47.
The updating unit 47, when can be used for after the completion of conversation shift or establishing new Dialog processing task, to institute
The session distribution information for stating data processing unit in data processor is updated.
The acquiring unit 41 specifically can be used for obtaining according to prefixed time interval in the data processor at data
The session for managing unit distributes information.And then it can be with the load balancing between data processing unit in real-time ensuring data processor.
It should be noted that its of each functional unit involved by another kind load balancing apparatus provided in an embodiment of the present invention
He accordingly describes, can be with reference to the corresponding description in Fig. 2, and details are not described herein.
Another kind load balancing apparatus provided in an embodiment of the present invention, can be only fitted in terminal device, can be used for controlling
Conversation shift between data processing unit processed, specifically includes: acquiring unit, migration units, recording unit, is divided at generation unit
With unit, copied cells, updating unit etc., the session of data processing unit in data processor is obtained by acquiring unit first
Distribute information；Then information and default load balance scheduling rule are distributed according to session by generation unit, generates data processing
The conversation dispatching information of load balancing between unit；Finally by migration units according to conversation dispatching information, in former data processing
The data accepted packet after treatment for the session that unit migrates needs, then the migration operation of the session is carried out, it may be implemented
Under the premise of not interrupting session traffic processing, the load balancing between different data processing units is realized, in conversation shift mistake
The problems such as Cheng Zhong, the data packet of corresponding session is not in packet loss, delay is excessive, synchronous, forwarding data flow is in transition process
In it is unaffected；It may be implemented to obtain the newest session distribution information of data processing unit in data processor in real time, and according to
The newest session distribution information executes conversation shift operation, and then can be with data processing unit in real-time ensuring data processor
Between load balancing, maximize the treatment effeciency of data processor.
Based on above-mentioned method and apparatus embodiment, the embodiment of the invention provides a kind of showing for SiteServer LBS framework
It is intended to, as shown in fig. 6, the system includes: distributor, data processor, scheduler, transference apparatus.Distributor may include can
By packet delivery to the function of different data processing units, it is specifically as follows the function of hardware view realization, such as part network interface card
RSS (Receive Side Scaling) function, FlowDirector function of hardware internal etc. is also possible to software view reality
Existing function, such as RPS (receive packet steering) function that linux kernel network module packet receiving mechanism provides,
Packet delivery is gone out using different classification methods by preset configuration parameter.Needle may be implemented in data processor
Processing to network packet main security business.Scheduler can be according to the session of data processing unit in data processor point
With information, the dispatching algorithm of load balancing is executed, generates the conversation dispatching information of load balancing between data processing unit；Migration
Device carries out the conversation shift in data processor between different data processing units according to the conversation dispatching information that scheduler generates,
So that the number of sessions of each data processing unit is balanced, and then load balancing between different data processing units may be implemented.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, reference can be made to the related descriptions of other embodiments.
It is understood that the correlated characteristic in the above method and device can be referred to mutually.In addition, in above-described embodiment
" first ", " second " etc. be and not represent the superiority and inferiority of each embodiment for distinguishing each embodiment.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
Algorithm and display are not inherently related to any particular computer, virtual system, or other device provided herein.
Various general-purpose systems can also be used together with teachings based herein.As described above, it constructs required by this kind of system
Structure be obvious.In addition, the present invention is also not directed to any particular programming language.It should be understood that can use various
Programming language realizes summary of the invention described herein, and the description done above to language-specific is to disclose this hair
Bright preferred forms.
In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that implementation of the invention
Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this specification.
Similarly, it should be understood that in order to simplify the disclosure and help to understand one or more of the various inventive aspects,
Above in the description of exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes
In example, figure or descriptions thereof.However, the disclosed method should not be interpreted as reflecting the following intention: i.e. required to protect
Shield the present invention claims features more more than feature expressly recited in each claim.More precisely, as following
Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
Thus the claims for following specific embodiment are expressly incorporated in the specific embodiment, wherein each claim itself
All as a separate embodiment of the present invention.
Those skilled in the art will understand that can be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more devices different from this embodiment.It can be the module or list in embodiment
Member or component are combined into a module or unit or component, and furthermore they can be divided into multiple submodule or subelement or
Sub-component.Other than such feature and/or at least some of process or unit exclude each other, it can use any
Combination is to all features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed
All process or units of what method or apparatus are combined.Unless expressly stated otherwise, this specification is (including adjoint power
Benefit require, abstract and attached drawing) disclosed in each feature can carry out generation with an alternative feature that provides the same, equivalent, or similar purpose
It replaces.
In addition, it will be appreciated by those of skill in the art that although some embodiments described herein include other embodiments
In included certain features rather than other feature, but the combination of the feature of different embodiments mean it is of the invention
Within the scope of and form different embodiments.For example, in the following claims, embodiment claimed is appointed
Meaning one of can in any combination mode come using.
Various component embodiments of the invention can be implemented in hardware, or to run on one or more processors
Software module realize, or be implemented in a combination thereof.It will be understood by those of skill in the art that can be used in practice
Microprocessor or digital signal processor (DSP) realize a kind of load-balancing method and device according to an embodiment of the present invention
In some or all components some or all functions.The present invention is also implemented as described herein for executing
Some or all device or device programs (for example, computer program and computer program product) of method.In this way
Realization program of the invention can store on a computer-readable medium, or can have the shape of one or more signal
Formula.Such signal can be downloaded from an internet website to obtain, and perhaps be provided on the carrier signal or with any other shape
Formula provides.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and ability
Field technique personnel can be designed alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between parentheses should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not
Element or step listed in the claims.Word "a" or "an" located in front of the element does not exclude the presence of multiple such
Element.The present invention can be realized by means of including the hardware of several different elements and by means of properly programmed computer.
In the unit claims listing several devices, several in these devices, which can be through the same hardware branch, has
Body embodies.The use of word first, second, and third does not indicate any sequence.These words can be construed to title.
Claims (12)
1. a kind of load-balancing method characterized by comprising
The session for obtaining data processing unit in data processor distributes information；
Information and default load balance scheduling rule are distributed according to the session, generates load balancing between data processing unit
Conversation dispatching information；
The conversation shift in the data processor between different data processing units is carried out according to the conversation dispatching information；
If the currently employed session Hash table of the data processor is local session Hash table, the method also includes:
The session information data field that will be migrated session copies to and described to be migrated session corresponding target data processing single
In the session Hash table of member.
2. load-balancing method according to claim 1, which is characterized in that include the number in the session distribution information
According to the currently processed meeting of each data processing unit in the number of data processing unit in processor and the data processor
Quantity is talked about, it is described to distribute information and default load balance scheduling rule according to the session, it generates and is born between data processing unit
Balanced conversation dispatching information is carried, is specifically included:
According to the number of sessions, the number of data processing unit and default load balance scheduling rule, determine at each data
The session and former data processing unit corresponding with the session for needing to migrate that need to migrate in reason unit and target data are handled
Unit；
The session migrated and former data processing unit corresponding with the session for needing to migrate and target data are needed according to described
Processing unit generates the conversation dispatching information of load balancing between data processing unit.
3. load-balancing method according to claim 1 or 2, which is characterized in that described according to the conversation dispatching information
The conversation shift in the data processor between different data processing units is carried out, is specifically included:
In the data accepted packet after treatment for the session that former data processing unit migrates needs, according to the conversation dispatching
Information, by the conversation shift for needing to migrate to target data processing unit.
4. load-balancing method according to claim 1, which is characterized in that described to be carried out according to the conversation dispatching information
After conversation shift in the data processor between different data processing units, the method also includes:
Record the identification information of the corresponding target data processing unit of session of migration；
When receiving data packet corresponding with the session of the migration, according to the identification information, by the allocation of packets
It is handled into the target data processing unit.
5. load-balancing method according to claim 1, which is characterized in that the method also includes:
After the completion of conversation shift or when establishing new Dialog processing task, to data processing unit in the data processor
Session distribution information is updated.
6. load-balancing method according to claim 5, which is characterized in that data processing in the acquisition data processor
The session of unit distributes information, specifically includes:
Information is distributed according to the session that prefixed time interval obtains data processing unit in the data processor.
7. a kind of load balancing apparatus characterized by comprising
Acquiring unit, the session for obtaining data processing unit in data processor distribute information；
Generation unit, session distribution information and default load balance scheduling rule for being obtained according to the acquiring unit are raw
At the conversation dispatching information of load balancing between data processing unit；
Migration units, the conversation dispatching information for being generated according to the generation unit carry out different numbers in the data processor
According to the conversation shift between processing unit；
Described device further include:
Copied cells will be by if the session Hash table currently employed for the data processor is local session Hash table
The session information data field of migration session copies to and the session for being migrated the corresponding target data processing unit of session
In Hash table.
8. load balancing apparatus according to claim 7, which is characterized in that include the number in the session distribution information
According to the currently processed meeting of each data processing unit in the number of data processing unit in processor and the data processor
Quantity is talked about, the generation unit specifically includes:
Determining module, for regular according to the number of the number of sessions, data processing unit and default load balance scheduling, really
The session and former data processing unit corresponding with the session for needing to migrate that need to migrate in fixed each data processing unit and
Target data processing unit；
Generation module, the session that migrates of needs for being determined according to the determining module and corresponding with the session that needs migrate
Former data processing unit and target data processing unit, generate data processing unit between load balancing conversation dispatching letter
Breath.
9. load balancing apparatus according to claim 7 or 8, which is characterized in that
The migration units, specifically for the data accepted packet processing knot of the session migrated in former data processing unit to needs
Shu Hou, according to the conversation dispatching information, by the conversation shift for needing to migrate to target data processing unit.
10. load balancing apparatus according to claim 7, which is characterized in that described device further include:
Recording unit, the identification information of the corresponding target data processing unit of session for recording migration；
Allocation unit, for when receiving data packet corresponding with the session of the migration, according to the recording unit records
Identification information, the allocation of packets is handled into the target data processing unit.
11. load balancing apparatus according to claim 7, which is characterized in that described device further include:
Updating unit, for after the completion of conversation shift or when establishing new Dialog processing task, in the data processor
The session distribution information of data processing unit is updated.
12. load balancing apparatus according to claim 7, which is characterized in that
The acquiring unit, specifically for obtaining the meeting of data processing unit in the data processor according to prefixed time interval
Words distribution information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610718285.1A CN106357726B (en) | 2016-08-24 | 2016-08-24 | Load-balancing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610718285.1A CN106357726B (en) | 2016-08-24 | 2016-08-24 | Load-balancing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106357726A CN106357726A (en) | 2017-01-25 |
| CN106357726B true CN106357726B (en) | 2019-08-20 |
Family
ID=57844960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610718285.1A Active CN106357726B (en) | 2016-08-24 | 2016-08-24 | Load-balancing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106357726B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108696557B (en) * | 2017-04-11 | 2021-08-10 | 北京京东尚科信息技术有限公司 | Information processing system, method and device |
| CN109714269B (en) * | 2018-12-19 | 2022-05-17 | 迈普通信技术股份有限公司 | Data processing method and network equipment |
| CN109495504B (en) * | 2018-12-21 | 2021-05-25 | 东软集团股份有限公司 | Firewall equipment and message processing method and medium thereof |
| CN110531931B (en) * | 2019-08-22 | 2022-03-22 | 济南浪潮数据技术有限公司 | Storage device selection method and device and computer readable storage medium |
| CN111628941A (en) * | 2020-05-27 | 2020-09-04 | 广东浪潮大数据研究有限公司 | Network traffic classification processing method, device, equipment and medium |
| CN112311876B (en) * | 2020-10-30 | 2022-08-16 | 武汉绿色网络信息服务有限责任公司 | Method and device for determining back-end device of current session through load balancing device |
| CN115150448B (en) * | 2022-06-14 | 2023-08-25 | 北京车网科技发展有限公司 | Session data processing method, system, storage medium and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102495857A (en) * | 2011-11-21 | 2012-06-13 | 北京新媒传信科技有限公司 | Load balancing method for distributed database |
| CN102662763A (en) * | 2012-04-11 | 2012-09-12 | 华中科技大学 | Virtual machine resource scheduling method based on service quality |
| CN103002038A (en) * | 2012-12-13 | 2013-03-27 | 华为技术有限公司 | Processing method and device for elastic load balancing |
| CN103259740A (en) * | 2013-04-27 | 2013-08-21 | 杭州华三通信技术有限公司 | Load balancing processing method and device |
| CN104901898A (en) * | 2015-06-08 | 2015-09-09 | 东软集团股份有限公司 | Load balancing method and device |
-
2016
- 2016-08-24 CN CN201610718285.1A patent/CN106357726B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102495857A (en) * | 2011-11-21 | 2012-06-13 | 北京新媒传信科技有限公司 | Load balancing method for distributed database |
| CN102662763A (en) * | 2012-04-11 | 2012-09-12 | 华中科技大学 | Virtual machine resource scheduling method based on service quality |
| CN103002038A (en) * | 2012-12-13 | 2013-03-27 | 华为技术有限公司 | Processing method and device for elastic load balancing |
| CN103259740A (en) * | 2013-04-27 | 2013-08-21 | 杭州华三通信技术有限公司 | Load balancing processing method and device |
| CN104901898A (en) * | 2015-06-08 | 2015-09-09 | 东软集团股份有限公司 | Load balancing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106357726A (en) | 2017-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106357726B (en) | Load-balancing method and device | |
| CN114189571B (en) | Apparatus and method for implementing accelerated network packet processing | |
| CN105721535B (en) | For carrying out calculating equipment, method and the machine readable storage medium of parallel processing to the service function in service function chain | |
| RU2643626C1 (en) | Method of distributing acceptable packages, queue selector, package processing device and information media | |
| US11394649B2 (en) | Non-random flowlet-based routing | |
| US20200007445A1 (en) | Enhanced service function chain | |
| US20190104022A1 (en) | Policy-based network service fingerprinting | |
| EP3407563B1 (en) | Method, apparatus and machine readable medium for accelerating network security monitoring | |
| US9910687B2 (en) | Data flow affinity for heterogenous virtual machines | |
| US10616099B2 (en) | Hypervisor support for network functions virtualization | |
| WO2015114473A1 (en) | Method and apparatus for locality sensitive hash-based load balancing | |
| Hu et al. | Towards" full containerization" in containerized network function virtualization | |
| CN112437023B (en) | Virtualized security network element data processing method, system, medium and cloud platform | |
| JP5900352B2 (en) | Packet processing apparatus, packet processing method and program | |
| CN106603409B (en) | Data processing system, method and equipment | |
| US20210034546A1 (en) | Transparent encryption | |
| US9491098B1 (en) | Transparent network multipath utilization through encapsulation | |
| US10616116B1 (en) | Network traffic load balancing using rotating hash | |
| Miguel-Alonso | A research review of OpenFlow for datacenter networking | |
| EP4199427A1 (en) | Ai-supported network telemetry using data processing unit | |
| US20160261526A1 (en) | Communication apparatus and processor allocation method for the same | |
| US9736080B2 (en) | Determination method, device and storage medium | |
| Chen et al. | Evolving switch architecture toward accommodating in-network intelligence | |
| CN106549815B (en) | Apparatus and method for real-time deep application recognition in a network | |
| Zhang et al. | Efficient and verifiable service function chaining in NFV: Current solutions and emerging challenges |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |