CN105897836A - Back source request processing method and device - Google Patents
Back source request processing method and device Download PDFInfo
- Publication number
- CN105897836A CN105897836A CN201510890400.9A CN201510890400A CN105897836A CN 105897836 A CN105897836 A CN 105897836A CN 201510890400 A CN201510890400 A CN 201510890400A CN 105897836 A CN105897836 A CN 105897836A
- Authority
- CN
- China
- Prior art keywords
- source
- response time
- server
- request
- source server
- 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.)
- Pending
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
- H04L67/50—Network services
- H04L67/56—Provisioning of proxy services
Abstract
The invention provides a back source request processing method and device. The method comprise the steps that a back source request currently initiated by a node server is monitored; according to the history response time of each source server responding to a history back source request, the trend response time of each source server for a current back source request is predicted, wherein the history response time is acquired by the node server; according to the trend response time of each source server for the current back source request, a target source server which is used for receiving the current back source request is selected; and the current back source request is sent to the target server source. According to the back source request processing method and device, by taking full account of the current processing capability of a source server, a back source server is selected; the back source request response speed of the node server is ensured; and the user experience of a user is ensured.
Description
Technical field
The present embodiments relate to network communication technology field, particularly relate to a kind of time source request processing method
And device.
Background technology
CDN (full name is CONTENT DELIVERY NETWORK, i.e. content distributing network)
It is by increasing by one layer of new network architecture in existing Internet, is published to data content connect most
The network " edge " (node server) of nearly user, makes subscription client can obtain required content nearby,
Shortening user and obtain the response time of data, (full name is CONTENT DELIVERY to CDN
NETWORK, i.e. content distributing network) network architecture comprises multiple source server and multiple node clothes
Business device, client is initiated data acquisition request to a node server, node server will be taken from source
The data feedback obtained in business device is to client.
And all data in source server can not be obtained due to node server, when in node server
Data needed for uncached user or data cached when having expired, in prior art, node server
Source request can be initiated back, with from the source server chosen to the source server of immediate or frequent connection
The corresponding data of middle acquisition are also handed down to subscription client.But when the source server that node server is chosen
Disposal ability poor time, by cause node server return source request response time long, and then impact
Consumer's Experience.
Summary of the invention
The embodiment of the present invention provides one to return source request processing method and device, in order to solve in prior art
The long technical problem of source of the returning request processing mode response time that causes back source to be asked.
The embodiment of the present invention provides one to return source request processing method, including:
Monitor source of the returning request that node server is currently initiated;
When each source server response history obtained according to described node server returns the historical responses of source request
Between, it was predicted that each source server is to the current trend response time returning source request;
According to each source server to the current trend response time returning source request, select to be used for receiving current returning
The target source server of source request;
To currently return source request to send to described target source server.
The embodiment of the present invention provides one to return source request processing method, including:
Monitor the current time source request that node server is initiated to the first source server;
The described first source server response history obtained according to described node server returns the history of source request
Response time, it was predicted that described first source server is to the current trend response time returning source request;
If described first source server exceeds, to the current trend response time returning source request, the response preset
Time threshold, returns going through of source request according to the second source server response history that described node server obtains
History response time, it was predicted that described second source server is to the current trend response time returning source request;
If the current trend response time returning source request is less than described default by described second source server
Response time threshold value, will currently return source request and send to described second source server.
The embodiment of the present invention provides one to return source request processing means, including:
Monitor module, for monitoring source of the returning request that node server is currently initiated;
Data processing module, returns for each source server response history obtained according to described node server
The historical responses time of source request, it was predicted that each source server is to the current trend response time returning source request;
Scheduling module, for returning, to current, the trend response time that source is asked according to each source server, selects
For receiving the current target source server returning source request;And, will currently return source request and send to described
Target source server.
The embodiment of the present invention provides one to return source request processing means, including:
Monitoring module, the current source of returning initiated to the first source server for monitoring node server is asked;
Data processing module, for the described first source server response obtained according to described node server
History returns the historical responses time of source request, it was predicted that current time source request is become by described first source server
Gesture response time;And,
If described first source server exceeds, to the current trend response time returning source request, the response preset
Time threshold, returns going through of source request according to the second source server response history that described node server obtains
History response time, it was predicted that described second source server is to the current trend response time returning source request;
Scheduling module, if for described second source server to the current trend response time returning source request
Less than described default response time threshold value, will currently return source request and send to described second source server.
What the embodiment of the present invention provided returns source request processing method and device, returns history according to source server
The historical responses time of source request, it was predicted that go out source server and return, to current, the trend response time that source is asked,
With reference to described trend response time, determine for responding the current target source server returning source request.?
In the case of taking into full account source server current processing capabilities, carry out back the selection of source server, thus can
Ensure node server returns source request answer speed.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to reality
Execute the required accompanying drawing used in example or description of the prior art to be briefly described, it should be apparent that under,
Accompanying drawing during face describes is some embodiments of the present invention, for those of ordinary skill in the art,
On the premise of not paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
Fig. 1 is that the present invention returns source request processing method flow chart;
Fig. 2 is the embodiment flow chart of anticipation trend response time in the embodiment of the present invention;
Fig. 3 is the embodiment flow chart selecting target source server in the embodiment of the present invention;
Fig. 4 is that the present invention returns source request processing method flow chart;
Fig. 5 is that the present invention returns source request processing means structural representation.
Detailed description of the invention
For making the purpose of the embodiment of the present invention, technical scheme and advantage clearer, below in conjunction with this
Accompanying drawing in bright embodiment, is clearly and completely described the technical scheme in the embodiment of the present invention,
Obviously, described embodiment is a part of embodiment of the present invention rather than whole embodiments.Based on
Embodiment in the present invention, those of ordinary skill in the art are obtained under not making creative work premise
The every other embodiment obtained, broadly falls into the scope of protection of the invention.
What the embodiment of the present invention provided returns source request processing method and device, and (full name is to can be applicable to CDN
CONTENT DELIVERY NETWORK, i.e. content distributing network) in framework.Node server
As CDN (full name is CONTENT DELIVERY NETWORK, i.e. content distributing network)
The carrier of distribution, data needed for user uncached in it or data cached when having expired, need
Source request is initiated back to source server.In prior art, node server return source request will send to by
CDN (full name is CONTENT DELIVERY NETWORK, i.e. content distributing network) webmaster
Center is pressed in the source server of nearby principle allotment, and after the source request of returning sends, node server needs to wait for source
The response of server, and the response time of source server can be by many-sides such as networking situation, load states
Impact, if the response time of source server is longer, the waiting time of node server also can be correspondingly
Extending, this will certainly affect user and obtain when the speed of data, such as user watch video and will result in video
Card, will result in the load time long etc., have impact on Consumer's Experience when checking webpage.
What the embodiment of the present invention provided returns source request processing method and device, be just intended to change existing by
CDN (full name is CONTENT DELIVERY NETWORK, i.e. content distributing network) webmaster
Center carries out source of the returning request processing mode of source server allotment, and is serviced according to source by node server side
The disposal ability that device is current, selects optimum to perform back the source server of source request, thus source of saving back please
The response time asked, and then ensure Consumer's Experience.
It addition, time source request processing method and device that the embodiment of the present invention provides, except can be applicable to
CDN (full name is CONTENT DELIVERY NETWORK, i.e. content distributing network) framework
In scene, apply also for other time source request and process in scene, do not limit at this.
With reference to Fig. 1, the embodiment of the present invention provides one to return source request processing method, including:
S101, monitors source of the returning request that node server is currently initiated;
S102, each source server response history obtained according to described node server returns the history of source request
Response time, it was predicted that each source server is to the current trend response time returning source request;
S103, according to each source server to the current trend response time returning source request, selects to be used for receiving
The current target source server returning source request;
S104, will currently return source request and send to described target source server.
Wherein, source of the returning request that step S101 interior joint server is currently initiated refers to: user is to node
When server request obtains data and there is currently no the data needed for user in node server, node
What server was actively initiated returns source request, and now, the sending object returning source request does not determines temporarily.
When listening to source of the returning request that node server is initiated, step S102 performs each source server
The prediction of trend response time.The response time of described time source request may be defined as: sends out from node server
Play back source request and be issued to, to returning source data, the time that node server is spent completely, naturally it is also possible to be
Other can reflect the source server timing statistics to the source of returning request disposal ability, the most specifically limits at this
Fixed.Described historical responses time, trend response time all refer to definition herein.
Node server, to there being at least one source server, preferably returns source request response to obtain,
Step S103 is using the trend response time of each source server as the reference selected, thus can select suitable
Close the target source server performing back source request.So far, it may be determined that go out current source of going back in step S101 and ask
The sending object asked, is target source server.
What the present embodiment provided returns source request processing method, please in the source of returning listening to node server initiation
When asking, there emerged a source server according to the historical responses time prediction of each source server corresponding to node server
Trend response time, and using the trend response time of each source server that dopes as foundation, determine
Go out to be suitably executed back the target source server of source request, and source of the returning request initiated by node server sends
To target source server.By predicting the service ability of each source server, can be determined by node server side
It is more suitable for performing back the source server of source request, thus when saving the response of source of returning request of node server
Between, it is achieved the data distributing obtained in the source of returning at faster speed is to user.
It should be noted that the prediction process in the present embodiment, to the trend response time of each source server
Be not limited to shown in Fig. 1 being positioned at monitor node server currently initiate return source request after, in advance
The scheme that survey process is additionally arranged at before monitoring source of the returning request that node server is currently initiated falls within this
The protection domain of bright embodiment.It addition, other step in the present embodiment is carried out also dependent on being actually needed
The adjustment of execution sequence, the technical scheme of the gained that thus develops falls within the protection domain of the embodiment of the present invention.
The mode predicting each source server trend response time to currently returning source request in step S102
Varied, with some embodiments, various prediction mode are described in detail below.
In one embodiment, each source server of described prediction returns, to current, the trend response time that source is asked,
Farther include:
Response time matched curve, and root is drawn respectively according to the historical responses time that each source server is corresponding
Predict that each source server is to the current trend response time returning source request according to described response time matched curve.
As a example by a source server, source server performs the source of repeatedly going back to of node server history transmission please
The historical responses time asked may be for discrete multiple data, can be by discrete many numbers by matched curve
Reacting on a curve of off-line data rule according to being fitted to approx, and the trend of moving towards according to curve can
Dope the trend response time next time returning source request.The present embodiment is that each source server is drawn respectively
Matched curve, and when predicting the trend response of each source server respectively according to the matched curve of each source server
Between.
In the present embodiment, matched curve can obtain by performing relevant many algorithms, and the present invention is to this also
It is not specifically limited.In order to obtain trend response time more accurately, it is preferable that Kalman can be used to filter
Ripple algorithm filters the burst response time in the historical responses time that source server is corresponding, going through after filtering
History response time matching generates response time matched curve;Wherein, described burst response time is: with
Before or the difference of posterior response time exceed the response time of predetermined value.
As described above, the historical responses time of source server is discrete multiple data, among these may
Can there is the bursty data the most excessive with other data difference, these bursty datas are likely due to source clothes
Business device meets with accident or network link meets with unexpected and causes, and these fortuitous events are all temporary transient,
Not being the normal operating conditions of source server, therefore, these bursty datas should not be used as prediction source service
The reference of the trend response time of device.In the present embodiment, by the historical responses time corresponding for source server
Burst response time filter, thus above-mentioned fortuitous event is got rid of, to avoid above-mentioned meaning
Predicting the outcome of outer situation effect tendency response time.
In another embodiment, with reference to Fig. 2, it was predicted that the current trend returning source request is rung by each source server
Process between Ying Shi, farther includes:
S1021, according to the Connecting quantity of each source server got, is converted to described Connecting quantity
Time delay response time;
S1022, the historical responses time corresponding according to described each source server and time delay response time, in advance
Survey each source server to the current trend response time returning source request;
Wherein, the Connecting quantity of described node server includes: connection failure rate, connective stability, negative
Carry one or more in quantity.
In the present embodiment, the Connecting quantity of each source server can carry out Tong Bu acquisition with its historical responses time,
That is to say at node server after source server initiates back source request, perform Hui Yuan obtaining source server
While the response time of request, obtain source server and perform the Connecting quantity during the request of this time time source,
For example whether connection failure, return in source procedure and whether interruption occurred, return the negative of source server in source procedure
Carry quantity etc..And Connecting quantity is converted to time delay response time, the strategy of conversion can be according to actual needs
It is set, if such as can set in this time time source procedure, interruption occurred, then this interruption is converted to
The time delay response time of 5s;Or this time is returned the load number of source server in source procedure and is alreadyd more than 5,
The time delay response time that each load transfer is 1s that then will exceed.
The sound that the time delay response time of source server and historical responses time sum will return source request as single
Answer data, and as the reference of anticipation trend response time.Such as, in foregoing embodiments, with matching
During the mode anticipation trend response time of curve, can be by the time delay response time of source server and historical responses
Time sum is as the data for drawing matched curve.Therefore, the present embodiment is considering source server pair
On the basis of the actual response time of the source of returning request, also consider and affected source server to the source of returning request
The various Connecting quantity of response time, and after described Connecting quantity is converted to time delay response time, will
Time delay response time and the historical responses time of described source server ring collectively as prediction source server trend
Foundation between Ying Shi, thus the prediction of trend response time more accurately can be realized.
With reference to Fig. 3, with an embodiment, step S103 is described in detail the most again.
In step S103, select the process for receiving the current target source server returning source request, enter
One step includes:
S1031, screening acquisition trend response time is without departing from the source server of default response time threshold value;
S1032, compares the trend response time of the described source server filtered out, and when trend is responded
Between the shortest source server be defined as target source server.
Wherein, described default response time threshold value can be set according to practical situation, such as peak in night
Default response time threshold value can be set as slightly higher value by the phase, to avoid due to each source corresponding to node server
The treating capacity of server is the biggest and the trend response time of causing all exceeds default response time threshold value, and then
The situation causing node server cannot obtain back the significant response that source is asked occurs.When meeting trend response
When time has multiple without departing from the source server of default response time threshold condition, further from eligible
Multiple source servers in pick out the shortest source server of trend response time as perform current source of going back to please
The target source server asked.
It addition, the prediction mode of trend response time can use arbitrary enforcement mentioned above in the present embodiment
The prediction mode that example provides, according to trend response time from multiple source servers that node server is corresponding
The source server selecting optimum performs back source request, thus, node server can be from mesh within the shortest time
Mark source server obtains data, and by data distributing to user, for user, will enjoy with
Fast speed obtains the experience of data from node server.
With CDN, (full name is CONTENT DELIVERY NETWORK, i.e. content distribution below
Network) embodiments of the invention carry out explained in detail as application scenarios by framework.
It is positioned at Pekinese user A and is arranged on Pekinese's node server one film of request to closest
During data, in Pekinese's node server and this cinematic data uncached, in this case, node takes
Business device can be predicted assessment to the current processing capabilities of relative 3 source servers A, B, C,
When the history to this node server according to 3 source servers A, B, C returns the historical responses of source request
Between dope the trend response time of 3 source servers A, B, C respectively, such as, it was predicted that go out 3 sources
Server A, the trend response time of B, C are respectively 5s, 3s and 8s, wherein the becoming of source server C
Gesture response time is beyond default response time threshold value 6s, and therefore, source server C will be rejected, joint
Point server can send currently returning source request to the shortest source server B of trend response time, or also
Can send to the source server A the most nearest with it.By this preferentially connect by the way of, can select
Being best suitable for performing node server and return the source server of source request, node server can service from source rapidly
Obtaining back source data in device, and be handed down to user A, user A can enjoy smoothness when watching this film
Viewing experience.
With reference to Fig. 4, the embodiment of the present invention also provides for back source request processing method, including:
S201, monitors the current time source request that node server is initiated to the first source server;
S202, returns source request according to the described first source server response history that described node server obtains
The historical responses time, it was predicted that described first source server is to the current trend response time returning source request;
S203, presets if the current trend response time returning source request is exceeded by described first source server
Response time threshold value, according to described node server obtain second source server response history return source please
The historical responses time asked, it was predicted that described second source server is to the current trend response time returning source request;
S204, if the current trend response time returning source request is less than described by described second source server
The response time threshold value preset, will currently return source request and send to described second source server.
Wherein, in step S201, the first source server can be randomly choosed by node server, it is possible to base
In the principle connected nearby or most frequently, by CDN, (full name is CONTENT DELIVERY
NETWORK, i.e. content distributing network) network center carries out allocating or carried out actively by node server
Select, other strategy also can be used to select, as long as can primarily determine that what egress server was initiated
The current first-selected source server returning source request, is not especially limited at this.
At node server while the first source server initiates current time source request, step S202 is held
The prediction to the current trend response time returning source request of row the first source server.If described the of prediction
The trend response time of one source server meets the requirement of default response time threshold value, then keep node clothes
Business device is connected with source of the returning request of described first source server, and continues executing with back the transmission of source data;As
The trend response time of described first source server of fruit prediction does not meets wanting of default response time threshold value
Ask, node server will be switched off with described first source server return source request be connected, from node serve
Other source server that device is relevant select the second source server to carry out the prediction of trend response time, if
The trend response time of the second source server meets the requirement of default response time threshold value, then by the second source
Server is as performing back the target source server that source is asked, if the trend of the second source server responds
Between do not meet the requirement of default response time threshold value, then continue from other source relevant to node server
Server select another source server carry out the prediction of trend response time, until matching trend response
Time meets the source server of default response time threshold value, and the source server that will match to is as execution
The target source server of the source of returning request.
It should be noted that in the present embodiment, the prediction mode of trend response time can use any of the above-described
There is provided in embodiment predicts each source server prediction side to the current trend response time returning source request
Case.
With reference to Fig. 5, the embodiment of the present invention provides one to return source request processing means, including:
Monitor module 11, for monitoring source of the returning request that node server is currently initiated;
Data processing module 12, for each source server response history obtained according to described node server
The historical responses time of the source of returning request, it was predicted that each source server is to the current trend response time returning source request;
Scheduling module 13, for returning the trend response time that source is asked, choosing according to each source server to current
Select for receiving the current target source server returning source request;And, will currently return source request and send to institute
State target source server.
Wherein, source of the returning request that monitoring module 11 interior joint server is currently initiated refers to: user is to node
When server request obtains data and there is currently no the data needed for user in node server, node
What server was actively initiated returns source request, and now, the sending object returning source request does not determines temporarily.
When listening to source of the returning request that node server is initiated, data processing module 12 performs to take each source
The prediction of business device trend response time.The response time of described time source request may be defined as: from node serve
Device is initiated back source request and is issued to, to returning source data, the time that node server is spent completely, the most also may be used
To be that other can reflect the source server timing statistics to the source of returning request disposal ability, do not make at this to have
Body limits.Described historical responses time, trend response time all refer to definition herein.
Node server, to there being at least one source server, preferably returns source request response to obtain,
Scheduling module 13 is using the trend response time of each source server as the reference selected, thus can select suitable
Close the target source server performing back source request.So far, it may be determined that go out to monitor current source of going back in module 11 and ask
The sending object asked, is target source server.
What the present embodiment provided returns source request processing means, please in the source of returning listening to node server initiation
When asking, there emerged a source server according to the historical responses time prediction of each source server corresponding to node server
Trend response time, and using the trend response time of each source server that dopes as foundation, determine
Go out to be suitably executed back the target source server of source request, and source of the returning request initiated by node server sends
To target source server.By predicting the service ability of each source server, can be determined by node server side
It is more suitable for performing back the source server of source request, thus when saving the response of source of returning request of node server
Between, it is achieved the data distributing obtained in the source of returning at faster speed is to user.
The each source server of prediction in data processing module 12 is to the current trend response time returning source request
Mode varied, with some embodiments, various prediction mode are described in detail below.
In one embodiment, described data processing module, it is further used for:
Response time matched curve, and root is drawn respectively according to the historical responses time that each source server is corresponding
Predict that each source server is to the current trend response time returning source request according to described response time matched curve.
As a example by a source server, source server performs the source of repeatedly going back to of node server history transmission please
The historical responses time asked may be for discrete multiple data, can be by discrete many numbers by matched curve
Reacting on a curve of off-line data rule according to being fitted to approx, and the trend of moving towards according to curve can
Dope the trend response time next time returning source request.The present embodiment is that each source server is drawn respectively
Matched curve, and when predicting the trend response of each source server respectively according to the matched curve of each source server
Between.
In the present embodiment, matched curve can obtain by performing relevant many algorithms, and the present invention is to this also
It is not specifically limited.In order to obtain trend response time more accurately, it is preferable that described data process mould
Block 12 is additionally operable to, and uses Kalman filtering algorithm to filter dashing forward in the historical responses time that source server is corresponding
Sending out response time, the historical responses time match after filtering generates response time matched curve;Wherein,
Described burst response time is: with when the difference of front or posterior response time exceedes the response of predetermined value
Between.
As described above, the historical responses time of source server is discrete multiple data, among these may
Can there is the bursty data the most excessive with other data difference, these bursty datas are likely due to source clothes
Business device meets with accident or network link meets with unexpected and causes, and these fortuitous events are all temporary transient,
Not being the normal operating conditions of source server, therefore, these bursty datas should not be used as prediction source service
The reference of the trend response time of device.In the present embodiment, by the historical responses time corresponding for source server
Burst response time filter, thus above-mentioned fortuitous event is got rid of, to avoid above-mentioned meaning
Predicting the outcome of outer situation effect tendency response time.
In another embodiment, described data processing module, it is further used for:
According to the Connecting quantity of each source server got, described Connecting quantity is converted to time delay response
Time;
The historical responses time corresponding according to described each source server and time delay response time, it was predicted that each source takes
Business device is to the current trend response time returning source request;
Wherein, the Connecting quantity of described node server includes: connection failure rate, connective stability, negative
Carry one or more in quantity.
In the present embodiment, the Connecting quantity of each source server can be carried out by monitoring mould with its historical responses time
Block 11 carries out synchronizing to obtain, and that is to say at node server after source server initiates back source request, is obtaining
While taking the response time that source server performs back source request, acquisition source server performs this time go back to source please
Seek the Connecting quantity of period, for example whether whether connection failure, time source procedure occurred interruption, Hui Yuan
During the load number etc. of source server.And Connecting quantity is converted to time delay response time, conversion
Strategy can be set according to actual needs, if such as can set in this time time source procedure, interruption occurred,
Then this interruption is converted to the time delay response time of 5s;Or the load of source server in source procedure is returned in this time
Quantity alreadys more than 5, then the time delay response time that each load transfer is 1s that will exceed.
The sound that the time delay response time of source server and historical responses time sum will return source request as single
Answer data, and as the reference of anticipation trend response time.Such as, in foregoing embodiments, at data
The time delay of source server during anticipation trend response time, can be responded in the way of matched curve by reason module 12
Time and historical responses time sum are as the data for drawing matched curve.Therefore, the present embodiment exists
On the basis of considering the source server actual response time to the source of returning request, also consider and affected source clothes
The business device various Connecting quantity to the response time of the source of returning request, and described Connecting quantity is converted to time delay
After response time, by the time delay response time of described source server and historical responses time collectively as prediction
The foundation of source server trend response time, thus the prediction of trend response time more accurately can be realized.
Hereinafter with an embodiment, scheduling module 13 is described in detail again.
Scheduling module 13, is further used for:
Screening acquisition trend response time is without departing from the source server of default response time threshold value;
The trend response time of the described source server relatively filtered out, and by the shortest for trend response time
Source server is defined as target source server.
Wherein, described default response time threshold value can be set according to practical situation, such as peak in night
Default response time threshold value can be set as slightly higher value by the phase, to avoid due to each source corresponding to node server
The treating capacity of server is the biggest and the trend response time of causing all exceeds default response time threshold value, and then
The situation causing node server cannot obtain back the significant response that source is asked occurs.When meeting trend response
When time has multiple without departing from the source server of default response time threshold condition, further from eligible
Multiple source servers in pick out the shortest source server of trend response time as perform current source of going back to please
The target source server asked.
It addition, the prediction mode of trend response time can use arbitrary enforcement mentioned above in the present embodiment
The prediction mode that example provides, according to trend response time from multiple source servers that node server is corresponding
The source server selecting optimum performs back source request, thus, node server can be from mesh within the shortest time
Mark source server obtains data, and by data distributing to user, for user, will enjoy with
Fast speed obtains the experience of data from node server.
The embodiment of the present invention also provides for a kind of source of going back to and asks processing means, including:
Monitor module 11, please to the current source of returning that the first source server is initiated for monitoring node server
Ask;
Data processing module 12, rings for described first source server obtained according to described node server
History is answered to return the historical responses time of source request, it was predicted that described first source server returns what source was asked to current
Trend response time;And,
If described first source server exceeds, to the current trend response time returning source request, the response preset
Time threshold, returns going through of source request according to the second source server response history that described node server obtains
History response time, it was predicted that described second source server is to the current trend response time returning source request;
Scheduling module 13, if when the current trend returning source request is responded by described second source server
Between less than described default response time threshold value, will currently return source request and send to described second source server.
Wherein, monitoring in module 11, the first source server can be randomly choosed by node server, it is possible to base
In the principle connected nearby or most frequently, by CDN, (full name is CONTENT DELIVERY
NETWORK, i.e. content distributing network) network center carries out allocating or carried out actively by node server
Select, other strategy also can be used to select, as long as can primarily determine that what egress server was initiated
The current first-selected source server returning source request, is not especially limited at this.
At node server while the first source server initiates current time source request, data processing module
12 perform the prediction to the current trend response time returning source request of first source server.If the institute of prediction
The trend response time stating the first source server meets the requirement of default response time threshold value, then allocate mould
Block 13 will keep node server to be connected with source of the returning request of described first source server, and continues executing with back
The transmission of source data;If the trend response time of described first source server of prediction does not meets default
The requirement of response time threshold value, scheduling module 13 will control node server and disconnect and described first source service
Source of the returning request of device connects, and data processing module 12 is then from other source server relevant to node server
Middle selection the second source server carries out the prediction of trend response time, if the trend of the second source server is rung
Meet the requirement of default response time threshold value between Ying Shi, then scheduling module 13 using the second source server as
Perform back the target source server of source request, if the trend response time of the second source server does not meets pre-
If the requirement of response time threshold value, then continue to select from other source server relevant to node server
Select another source server and carry out the prediction of trend response time, until the trend response time of matching meets pre-
If the source server of response time threshold value, and the source server that will match to is as performing back source request
Target source server.
It should be noted that in the present embodiment, the prediction mode of trend response time can use any of the above-described
There is provided in embodiment predicts each source server prediction side to the current trend response time returning source request
Case.
Device embodiment described above is only schematically, wherein said illustrates as separating component
Unit can be or may not be physically separate, the parts shown as unit can be or
Person may not be physical location, i.e. may be located at a place, or can also be distributed to multiple network
On unit.Some or all of module therein can be selected according to the actual needs to realize the present embodiment
The purpose of scheme.Those of ordinary skill in the art are not in the case of paying performing creative labour, the most permissible
Understand and implement.
Through the above description of the embodiments, those skilled in the art is it can be understood that arrive each reality
The mode of executing can add the mode of required general hardware platform by software and realize, naturally it is also possible to by firmly
Part.Based on such understanding, the portion that prior art is contributed by technique scheme the most in other words
Dividing and can embody with the form of software product, this computer software product can be stored in computer can
Read in storage medium, such as ROM/RAM, magnetic disc, CD etc., including some instructions with so that one
Computer equipment (can be personal computer, server, or the network equipment etc.) performs each to be implemented
The method described in some part of example or embodiment.
Last it is noted that above example is only in order to illustrate technical scheme, rather than to it
Limit;Although the present invention being described in detail with reference to previous embodiment, the ordinary skill of this area
Personnel it is understood that the technical scheme described in foregoing embodiments still can be modified by it, or
Person carries out equivalent to wherein portion of techniques feature;And these amendments or replacement, do not make corresponding skill
The essence of art scheme departs from the spirit and scope of various embodiments of the present invention technical scheme.
Claims (12)
1. one kind is returned source request processing method, it is characterised in that including:
Monitor source of the returning request that node server is currently initiated;
When each source server response history obtained according to described node server returns the historical responses of source request
Between, it was predicted that each source server is to the current trend response time returning source request;
According to each source server to the current trend response time returning source request, select to be used for receiving current returning
The target source server of source request;
To currently return source request to send to described target source server.
Method the most according to claim 1, it is characterised in that each source server of described prediction is to working as
The front trend response time returning source request, farther includes:
According to the Connecting quantity of each source server got, described Connecting quantity is converted to time delay response
Time;
The historical responses time corresponding according to described each source server and time delay response time, it was predicted that each source takes
Business device is to the current trend response time returning source request;
Wherein, the Connecting quantity of described node server includes: connection failure rate, connective stability, negative
Carry one or more in quantity.
Method the most according to claim 1, it is characterised in that each source server of described prediction is to working as
The front trend response time returning source request, farther includes:
Response time matched curve, and root is drawn respectively according to the historical responses time that each source server is corresponding
Predict that each source server is to the current trend response time returning source request according to described response time matched curve.
Method the most according to claim 3, it is characterised in that according to corresponding the going through of each source server
History response time draws response time matched curve respectively, farther includes:
When using Kalman filtering algorithm to filter the burst response in the historical responses time that source server is corresponding
Between, the historical responses time match after filtering generates response time matched curve;
Wherein, described burst response time is: exceed predetermined with the difference at front or posterior response time
The response time of value.
Method the most according to claim 1, it is characterised in that return current according to each source server
The trend response time of source request, selects, for receiving the current target source server returning source request, to enter one
Step includes:
Screening acquisition trend response time is without departing from the source server of default response time threshold value;
The trend response time of the described source server relatively filtered out, and by the shortest for trend response time
Source server is defined as target source server.
6. one kind is returned source request processing method, it is characterised in that including:
Monitor the current time source request that node server is initiated to the first source server;
The described first source server response history obtained according to described node server returns the history of source request
Response time, it was predicted that described first source server is to the current trend response time returning source request;
If described first source server exceeds, to the current trend response time returning source request, the response preset
Time threshold, returns going through of source request according to the second source server response history that described node server obtains
History response time, it was predicted that described second source server is to the current trend response time returning source request;
If the current trend response time returning source request is less than described default by described second source server
Response time threshold value, will currently return source request and send to described second source server.
7. one kind is returned source request processing means, it is characterised in that including:
Monitor module, for monitoring source of the returning request that node server is currently initiated;
Data processing module, returns for each source server response history obtained according to described node server
The historical responses time of source request, it was predicted that each source server is to the current trend response time returning source request;
Scheduling module, for returning, to current, the trend response time that source is asked according to each source server, selects
For receiving the current target source server returning source request;And, will currently return source request and send to described
Target source server.
Device the most according to claim 7, it is characterised in that described data processing module, enters one
Step is used for:
According to the Connecting quantity of each source server got, described Connecting quantity is converted to time delay response
Time;
The historical responses time corresponding according to described each source server and time delay response time, it was predicted that each source takes
Business device is to the current trend response time returning source request;
Wherein, the Connecting quantity of described node server includes: connection failure rate, connective stability, negative
Carry one or more in quantity.
Device the most according to claim 7, it is characterised in that described data processing module, enters one
Step is used for:
Response time matched curve, and root is drawn respectively according to the historical responses time that each source server is corresponding
Predict that each source server is to the current trend response time returning source request according to described response time matched curve.
Device the most according to claim 9, it is characterised in that described data processing module, enters
One step is used for:
When using Kalman filtering algorithm to filter the burst response in the historical responses time that source server is corresponding
Between, the historical responses time match after filtering generates response time matched curve;
Wherein, described burst response time is: exceed predetermined with the difference at front or posterior response time
The response time of value.
11. devices according to claim 7, it is characterised in that described scheduling module, further
For:
Screening acquisition trend response time is without departing from the source server of default response time threshold value;
The trend response time of the described source server relatively filtered out, and by the shortest for trend response time
Source server is defined as target source server.
12. 1 kinds are returned source request processing means, it is characterised in that including:
Monitoring module, the current source of returning initiated to the first source server for monitoring node server is asked;
Data processing module, for the described first source server response obtained according to described node server
History returns the historical responses time of source request, it was predicted that current time source request is become by described first source server
Gesture response time;And,
If described first source server exceeds, to the current trend response time returning source request, the response preset
Time threshold, returns going through of source request according to the second source server response history that described node server obtains
History response time, it was predicted that described second source server is to the current trend response time returning source request;
Scheduling module, if for described second source server to the current trend response time returning source request
Less than described default response time threshold value, will currently return source request and send to described second source server.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510890400.9A CN105897836A (en) | 2015-12-07 | 2015-12-07 | Back source request processing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510890400.9A CN105897836A (en) | 2015-12-07 | 2015-12-07 | Back source request processing method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105897836A true CN105897836A (en) | 2016-08-24 |
Family
ID=57002915
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510890400.9A Pending CN105897836A (en) | 2015-12-07 | 2015-12-07 | Back source request processing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105897836A (en) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106357783A (en) * | 2016-09-29 | 2017-01-25 | 北京奇艺世纪科技有限公司 | Method and device for distributing edge nodes |
CN106375471A (en) * | 2016-09-29 | 2017-02-01 | 北京奇艺世纪科技有限公司 | Edge node determination method and device |
CN106550019A (en) * | 2016-10-20 | 2017-03-29 | 福建天泉教育科技有限公司 | Browse request processing method and system |
CN106656674A (en) * | 2016-12-29 | 2017-05-10 | 北京爱奇艺科技有限公司 | Data sourcing scheduling method and device |
CN107317879A (en) * | 2017-08-02 | 2017-11-03 | 网宿科技股份有限公司 | The distribution method and system of a kind of user's request |
CN107613025A (en) * | 2017-10-31 | 2018-01-19 | 武汉光迅科技股份有限公司 | A kind of implementation method replied based on message queue order and device |
CN107707938A (en) * | 2017-10-11 | 2018-02-16 | 湖南机友科技有限公司 | Reduce the method and device of live video playout-delay |
CN108390955A (en) * | 2018-05-09 | 2018-08-10 | 网宿科技股份有限公司 | Domain Name acquisition method, Website access method and server |
CN108810052A (en) * | 2017-05-03 | 2018-11-13 | 华为技术有限公司 | The selection method and device of direct broadcast server |
CN109600415A (en) * | 2018-10-23 | 2019-04-09 | 平安科技(深圳)有限公司 | The method, apparatus and computer equipment of target data are obtained from multiple source servers |
CN110099134A (en) * | 2018-09-26 | 2019-08-06 | 深圳市网心科技有限公司 | Video file method for down loading, device, electronic equipment and storage medium |
CN110311987A (en) * | 2019-07-24 | 2019-10-08 | 中南民族大学 | Node scheduling method, apparatus, equipment and the storage medium of microserver |
CN110636104A (en) * | 2019-08-07 | 2019-12-31 | 咪咕视讯科技有限公司 | Resource request method, electronic device and storage medium |
CN110719273A (en) * | 2019-09-29 | 2020-01-21 | 咪咕视讯科技有限公司 | Method for determining back source node, server and computer readable storage medium |
CN110825525A (en) * | 2019-11-06 | 2020-02-21 | 湖南快乐阳光互动娱乐传媒有限公司 | Data resource back-source method and device |
CN111052104A (en) * | 2017-08-25 | 2020-04-21 | 国际商业机器公司 | Server request management |
CN112653736A (en) * | 2020-12-10 | 2021-04-13 | 北京金山云网络技术有限公司 | Parallel source returning method and device and electronic equipment |
CN112866060A (en) * | 2021-01-25 | 2021-05-28 | 湖南快乐阳光互动娱乐传媒有限公司 | Server response time obtaining method and device |
CN113132133A (en) * | 2019-12-31 | 2021-07-16 | 中国移动通信集团陕西有限公司 | Method, device, computing equipment and storage medium for distributing user configuration data |
CN117555309A (en) * | 2024-01-12 | 2024-02-13 | 北京中航泰瑞电子技术有限公司 | Multi-way switch control optimization method and system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102368776A (en) * | 2011-11-25 | 2012-03-07 | 中国科学技术大学 | Optimization function module of node list in content distribution/delivery network (CDN) |
CN102868935A (en) * | 2012-08-24 | 2013-01-09 | 乐视网信息技术(北京)股份有限公司 | Scheduling method for responding multiple sources in content distribution network (CDN) |
CN102970381A (en) * | 2012-12-21 | 2013-03-13 | 网宿科技股份有限公司 | Multi-source load balance method and system for proportional polling based on content distribution network |
CN103546590A (en) * | 2013-10-18 | 2014-01-29 | 北京奇虎科技有限公司 | Method and device for choosing DNS (domain name server) |
US20150089352A1 (en) * | 2013-09-25 | 2015-03-26 | Akamai Technologies, Inc. | Key Resource Prefetching Using Front-End Optimization (FEO) Configuration |
-
2015
- 2015-12-07 CN CN201510890400.9A patent/CN105897836A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102368776A (en) * | 2011-11-25 | 2012-03-07 | 中国科学技术大学 | Optimization function module of node list in content distribution/delivery network (CDN) |
CN102868935A (en) * | 2012-08-24 | 2013-01-09 | 乐视网信息技术(北京)股份有限公司 | Scheduling method for responding multiple sources in content distribution network (CDN) |
CN102970381A (en) * | 2012-12-21 | 2013-03-13 | 网宿科技股份有限公司 | Multi-source load balance method and system for proportional polling based on content distribution network |
US20150089352A1 (en) * | 2013-09-25 | 2015-03-26 | Akamai Technologies, Inc. | Key Resource Prefetching Using Front-End Optimization (FEO) Configuration |
CN103546590A (en) * | 2013-10-18 | 2014-01-29 | 北京奇虎科技有限公司 | Method and device for choosing DNS (domain name server) |
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106375471A (en) * | 2016-09-29 | 2017-02-01 | 北京奇艺世纪科技有限公司 | Edge node determination method and device |
CN106357783A (en) * | 2016-09-29 | 2017-01-25 | 北京奇艺世纪科技有限公司 | Method and device for distributing edge nodes |
CN106357783B (en) * | 2016-09-29 | 2019-06-14 | 北京奇艺世纪科技有限公司 | A kind of fringe node distribution method and device |
CN106550019A (en) * | 2016-10-20 | 2017-03-29 | 福建天泉教育科技有限公司 | Browse request processing method and system |
CN106656674A (en) * | 2016-12-29 | 2017-05-10 | 北京爱奇艺科技有限公司 | Data sourcing scheduling method and device |
CN108810052A (en) * | 2017-05-03 | 2018-11-13 | 华为技术有限公司 | The selection method and device of direct broadcast server |
CN108810052B (en) * | 2017-05-03 | 2020-09-04 | 华为技术有限公司 | Live broadcast server selection method and device |
CN107317879A (en) * | 2017-08-02 | 2017-11-03 | 网宿科技股份有限公司 | The distribution method and system of a kind of user's request |
CN107317879B (en) * | 2017-08-02 | 2019-07-05 | 网宿科技股份有限公司 | A kind of distribution method and system of user's request |
CN111052104B (en) * | 2017-08-25 | 2023-12-19 | 国际商业机器公司 | Server request management |
CN111052104A (en) * | 2017-08-25 | 2020-04-21 | 国际商业机器公司 | Server request management |
CN107707938A (en) * | 2017-10-11 | 2018-02-16 | 湖南机友科技有限公司 | Reduce the method and device of live video playout-delay |
CN107707938B (en) * | 2017-10-11 | 2020-12-08 | 湖南机友科技有限公司 | Method and device for reducing live video playing delay |
CN107613025A (en) * | 2017-10-31 | 2018-01-19 | 武汉光迅科技股份有限公司 | A kind of implementation method replied based on message queue order and device |
CN108390955A (en) * | 2018-05-09 | 2018-08-10 | 网宿科技股份有限公司 | Domain Name acquisition method, Website access method and server |
CN110099134A (en) * | 2018-09-26 | 2019-08-06 | 深圳市网心科技有限公司 | Video file method for down loading, device, electronic equipment and storage medium |
CN110099134B (en) * | 2018-09-26 | 2022-04-01 | 深圳市网心科技有限公司 | Video file downloading method and device, electronic equipment and storage medium |
CN109600415A (en) * | 2018-10-23 | 2019-04-09 | 平安科技(深圳)有限公司 | The method, apparatus and computer equipment of target data are obtained from multiple source servers |
CN110311987A (en) * | 2019-07-24 | 2019-10-08 | 中南民族大学 | Node scheduling method, apparatus, equipment and the storage medium of microserver |
CN110636104A (en) * | 2019-08-07 | 2019-12-31 | 咪咕视讯科技有限公司 | Resource request method, electronic device and storage medium |
CN110719273A (en) * | 2019-09-29 | 2020-01-21 | 咪咕视讯科技有限公司 | Method for determining back source node, server and computer readable storage medium |
CN110825525A (en) * | 2019-11-06 | 2020-02-21 | 湖南快乐阳光互动娱乐传媒有限公司 | Data resource back-source method and device |
CN110825525B (en) * | 2019-11-06 | 2022-06-07 | 湖南快乐阳光互动娱乐传媒有限公司 | Data resource back-source method and device |
CN113132133A (en) * | 2019-12-31 | 2021-07-16 | 中国移动通信集团陕西有限公司 | Method, device, computing equipment and storage medium for distributing user configuration data |
CN113132133B (en) * | 2019-12-31 | 2023-08-15 | 中国移动通信集团陕西有限公司 | Method, device, computing equipment and storage medium for distributing user configuration data |
CN112653736A (en) * | 2020-12-10 | 2021-04-13 | 北京金山云网络技术有限公司 | Parallel source returning method and device and electronic equipment |
CN112866060A (en) * | 2021-01-25 | 2021-05-28 | 湖南快乐阳光互动娱乐传媒有限公司 | Server response time obtaining method and device |
CN112866060B (en) * | 2021-01-25 | 2022-05-17 | 湖南快乐阳光互动娱乐传媒有限公司 | Server response time obtaining method and device |
CN117555309A (en) * | 2024-01-12 | 2024-02-13 | 北京中航泰瑞电子技术有限公司 | Multi-way switch control optimization method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105897836A (en) | Back source request processing method and device | |
CN105052080B (en) | For providing method, system and the computer-readable medium of thinking diameter network framework | |
CN106452958B (en) | Flow control method, system and centralized controller | |
EP2913978A1 (en) | Business scheduling method and apparatus and convergence device | |
US10993127B2 (en) | Network slice instance management method, apparatus, and system | |
CN105335231B (en) | server thread dynamic allocation method and device | |
CN110505155B (en) | Request degradation processing method and device, electronic equipment and storage medium | |
CN109672711B (en) | Reverse proxy server Nginx-based http request processing method and system | |
EP3547625B1 (en) | Method and system for sending request for acquiring data resource | |
EP3465985B1 (en) | A method and system for distributed control of large photonic switched networks | |
CN103905337A (en) | Network resource processing device, method and system | |
KR20170126408A (en) | Apparatus and method for managing network automatically | |
CN106034153A (en) | Bandwidth resource distribution method, data transmission method, data transmission device and data processing system | |
CN103124436B (en) | A kind of wireless telecommunication system, cloud virtual base station and resource regulating method | |
CN110515728B (en) | Server scheduling method and device, electronic equipment and machine-readable storage medium | |
WO2011024930A1 (en) | Content distribution system, content distribution method and content distribution-use program | |
EP2849389A2 (en) | Method and apparatus for allocating bandwidth resources | |
CN105450540A (en) | Load balancing method and device as well as DHCP (dynamic host configuration protocol) server | |
KR101883671B1 (en) | Method and management server for dtitributing node | |
CN110995802A (en) | Task processing method and device, storage medium and electronic device | |
JP5829230B2 (en) | Management system and management method | |
EP3435615B1 (en) | Network service implementation method, service controller, and communication system | |
CN115604263A (en) | Resource scheduling method and device, electronic equipment and storage medium | |
CN109831467A (en) | Data transmission method, equipment and system | |
CN113535402A (en) | Load balancing processing method and device based on 5G MEC and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160824 |