CN103200273A - Content replica consistency maintaining method based on physical location - Google Patents
Content replica consistency maintaining method based on physical location Download PDFInfo
- Publication number
- CN103200273A CN103200273A CN2013101372312A CN201310137231A CN103200273A CN 103200273 A CN103200273 A CN 103200273A CN 2013101372312 A CN2013101372312 A CN 2013101372312A CN 201310137231 A CN201310137231 A CN 201310137231A CN 103200273 A CN103200273 A CN 103200273A
- Authority
- CN
- China
- Prior art keywords
- node
- replica
- replica node
- copy
- responsibility
- 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.)
- Granted
Links
Images
Abstract
The invention discloses a content replica consistency maintaining method based on a physical location of a CDN-P2P (Content Distribution Network) (Peer-to-Peer) fusion content distribution network. The method comprises the following steps of: dividing a replica server into a plurality of groups according to the physical location, wherein nodes in the groups are close to each other in physical location; and establishing a replica set in each group, in order to spread updated messages in the replica set in a broadcasting way. The replica is selectively updated while updating, in order to further decrease the transmission amount of the updated messages, which means that only the replica nodes which are frequently accessed are updated. Voting is carried out in the replica set before updating each time, in order to avoid the conflict occurring in updating of the replica, and the updating cannot be carried out only when the amount of votes is more than half of the amount of the replica nodes in the groups. The content replica consistency maintaining method specifically comprises the following links of: dividing the replica server into node groups, establishing the replica set, maintaining the replica set, etc. By adopting the content replica consistency maintaining method, the delay of message transmission in the network can be effectively reduced, and the amount of redundant updating messages can be decreased.
Description
Technical field
The present invention relates to communication technical field, be specifically related to the method that CDN-P2P merges copy of content consistency maintenance in the content distributing network replica server and avoids the copy update inconsistency.
Background technology
Along with the Internet and Development of Multimedia Technology, streaming media video is increasing, has become the main bearing mode of internet information.This not only makes Virtual network operator and content supplier obtain a large amount of economic benefits, has also made things convenient for daily life simultaneously.
In traditional Web content distribution technology, CDN (CDN:Content Distribution Network) technology and P2P (P2P:Peer-to-Peer) technology are two big mainstream content distribution technologies.CDN is distributed in network edge to application server, adopts redirecting technique that user's request is turned to from the nearest Edge Server of user distance.The CDN technology has the advantage of high manageability and reliability, but its construction cost height and poor expandability.The P2P technology need not server, and it has taken full advantage of the resource of client, also can satisfy the requirement that the user increases sharply when not increasing cost, but the P2P node is not easy to manage and since node capacity restricted can not provide as the stable service of server.
It is a kind of network that is combined with P2P by CDN that CDN-P2P merges content distributing network, and its structure is divided into three layers: the CDN central server layer that the server that ground floor is provided by the content provider of website is formed, and it is top to be positioned at network; The second layer is formed by being distributed in each regional Edge Server, and these Edge Servers are also referred to as replica server.Adopting this structural reason is the replica server that takes full advantage of distribution various places under the existing client/server network structure.These servers receive the content from the upper strata central server, and provide service for lower floor's user node; The 3rd layer is the P2P user that is made of subscriber computer bunch.Each user bunch is made of with the P2P pattern the subscriber computer in the different area of a plurality of scales.As shown in Figure 1.The backbone network of this network has kept traditional CDN pattern, adopts the P2P pattern between the user under the replica server.CDN-P2P merges content distributing network not only the two advantage of CDN and P2P, and has overcome the deficiency of the two, therefore becomes present comparatively feasible content distributing network structure, is the focus of distributing network research field in recent years.
Merging content distributing network at CDN-P2P uses the copy technology effectively to improve utilance and the minimizing query responding time of network extensibility, resource.Along with the appearance that new network is used, shared resource frequent updating.For guaranteeing the true(-)running of system, effectively copy consistency maintaining method becomes key.
Merge content distributing network at CDN-P2P, the transmission quantity that how to reduce redundant updating message has become a major issue of needing solution badly.Because user's bunch copy content maintenance expense is big, do not have and safeguard necessaryly, in this maintenance is the copy of content consistency of replica server.At present, the copy of content consistency maintaining method is mainly taked following several thinking in the replica server:
1. transmit the message mode maintaining coherency by neighbours: updating message is spread all in the network by the mode that is transmitted to all or part neighbor node;
2. upgrade based on the consistency of copy tree: set up the n fork tree that is made of all replica node when transmitting updating message with neighbor node for the first time, later updating message is all only carried out in this " tree ".
Yet the shortcoming of said method thinking is: (1) redundant updating message is too much, the waste bandwidth resource; (2) do not consider that physical topology and logical topology do not match, waste bandwidth resource not only, and influence message propagation speed; (3) the copy access frequency that has is low, does not have necessity of renewal, and the renewal of copy is optionally carried out frequent accessed copy content, can further reduce the propagation of updating message.If the node division that physical location is close is in close identical group, and the copy set that foundation is made of replica node in the group, only upgrades and in copy set, carry out.This not only reduces the transmission of messages time-delay, also reduces redundant updating message simultaneously.For further alleviating the traffic, can delete the low replica node of access frequency in the copy set.
Summary of the invention
To merge the transmission quantity of the redundant updating message of content distributing network replica server layer too much in order to solve CDN-P2P, the technical problem that the updating message propagation delay is bigger, the present invention has overcome and has not considered physical topology in the network in the prior art and logical topology does not match, redundant updating message many and transmission delay is big defective, a kind of copy of content consistency maintaining method has been proposed, reducing the time-delay of transmission of messages as much as possible, and reduce the transmission of unnecessary updating message.
The present invention proposes the copy of content C consistency maintaining method based on physical location that a kind of CDN-P2P merges content distributing network, described CDN-P2P merges content distributing network and comprises central server layer, replica server layer and client layer, it is characterized in that, may further comprise the steps:
The responsibility replica node KN of step 3, each group
CiObtain place group G by the propagation of updating message
iThe replica node N of middle content copy C
cThe feedback message of sending that comprises its address is set up the copy set P of copy of content C
Ci, i=1,2 ..., L, and with described copy set P
CiInformation sends to copy set P
CiEach interior common replica node ON
Ci, i=1,2 ..., L, wherein, common replica node ON
CiBe copy set P
CiIn other replica node N except the responsibility replica node
c
Step 4, the utilization propagation of the updating message of copy of content C are for the first time set T with the responsibility replica node
cWith copy set P
CiAfter setting up well, when the updating message of meaningful copy C arrives again, from the responsibility replica node tree T of copy of content C
cRoot node N
c 1Beginning is successively transmitted, responsibility replica node tree T
cIn each responsibility replica node KN
CiReceive the copy set P that is broadcast to the place after the described updating message more respectively
CiIn.
Preferably, responsibility replica node KN
CiBe group G
iIn first receives the non-marked node of updating message.
Preferably, the mode of setting up a plurality of groups centered by flag node is: the group of the flag node that non-marked node chosen distance is nearest adds, among each group the node number less than threshold value M, and in arbitrary node and the group average distance of other node less than threshold value C.
Preferably, group G
iIn this group mark node M of each node periodic test N
iWhether effective, first determines this group mark node M N
iInvalid node updates is new flag node MN
i
Preferably, also comprise copy set P after the step 4
CiIn replica node leave safeguard step: when the replica node of leaving is responsibility replica node KN
CiThe time, responsibility replica node KN
CiSpecify new responsibility replica node in advance, and to responsibility replica node tree T
cWith copy set P
CiInterior each common replica node ON
CiSend leave group message; Be common replica node ON when leaving replica node
CiThe time, common replica node ON
CiIn advance to copy set P
CiInterior other replica node sends leave group message.
Preferably, also comprise copy set P after the step 4
CiIn replica node add safeguard step: the node that desire adds is to copy set P
CiAn interior replica node sends the message that joins request, and receives that the replica node of the described message that joins request contains place copy set P to the node transmission that described desire adds
CiThe feedback message of information, and to copy set P
CiInterior other replica node sends newcomer and adds message.
Preferably, also comprise copy set P after the step 4
CiIn replica node lost efficacy safeguard step: at each copy set P
CiBefore the described updating message of middle broadcasting, responsibility replica node KN
CiTo place copy set P
CiIn each common replica node ON
CiSend the request updating message, receive each common replica node ON
CiResponse message, and will not have the common replica node ON of response message
CiBe labeled as inefficacy.
Common replica node ON
CiPeriodic test place copy set P
CiResponsibility replica node KN
Ci, if responsibility replica node KN
CiLost efficacy, first finds the common replica node ON that it leaves
CiBe new responsibility replica node, the responsibility replica node that new responsibility replica node contacts other copy set adds responsibility replica node tree T again
c
Preferably, copy set P is set
CiThe node visit frequency threshold, as copy set P
CiWhen the node visit frequency of interior replica node was lower than described threshold value, described replica node was from copy set P
CiInterior deletion, and the copy of content C of the described replica node of mark is invalid.
Preferably, at each copy set P
CiBefore the interior broadcasting updating message, responsibility replica node KN
CiTo place copy set P
CiIn all nodes send request updating message, as request updating message number of responses WQ〉WQ node broadcasts to response during 1/2*N send updating message, the node that does not have response is considered as leaving copy set, wherein, N is crowd G
iThe middle effective replica node number of copy of content.
When non-replica node visit copy set P is arranged again
CiIn copy of content the time, described non-replica node is to copy set P
CiIn replica node initiate to read request message, when reading request message number of responses RQ〉during N-WQ, described non-replica node reads the highest copy of content C of version from RQ replica node of response.
The present invention proposes CDN-P2P and merges in the content distributing network copy of content consistency maintaining method based on physical location, this method has solved defective of the prior art: the one, and redundant updating message is many, the 2nd, do not consider the physical location of node, the 3rd, do not consider whether the low replica node of access frequency has necessity of renewal etc.Basic ideas of the present invention are that each replica server is divided into a plurality of groups, and the physical location of node is close among the group, sets up copy set by the responsibility replica node in the group, also set up a responsibility replica node tree simultaneously between each responsibility replica node.When updating message arrives, in responsibility replica node tree, propagate updating message earlier, after receiving, each responsibility replica node is broadcast to respectively in each copy set.In order further to reduce the quantity that updating message is propagated, with unwelcome replica node deletion in the copy set.In the copy renewal process, update inconsistency may occur, for anti-phenomenon here takes place, adopt the ballot mode to solve.The method can effectively reduce the transmission delay of updating message, and redundancy message quantity also can significantly reduce simultaneously.
Description of drawings
Fig. 1 CDN-P2P of the present invention merges the content distributing network structure chart.
Fig. 2 CDN-P2P of the present invention merges replica server layer division group schematic diagram in the content distributing network.
Copy set sets up procedure chart in Fig. 3 group of the present invention.
Build the structure chart behind the copy set among Fig. 4 group of the present invention.
Embodiment
For the course of work of the present invention can be described better, below in conjunction with accompanying drawing its specific implementation process is elaborated.
Fig. 1 is that CDN-P2P merges the content distributing network structure chart.Its structure is divided into three layers: server layer centered by the ground floor; The second layer is the replica server layer, adopts the P2P pattern between the server of this layer; The 3rd layer is P2P user's bunch layer.Replica server can obtain content resource to central server, provides service for user's bunch layer then.When having renewal in certain, certain the replica server node that contains this content is issued the place with other replica node of layer with this message after central server obtains updating message.Before sending out updating message, at first the node with the replica server layer is divided into a plurality of groups by physical location, and the node physical location is close among the group, sets up copy set in each group, and the copy updating message is carried out in the mode of broadcasting in collection.Node as far as possible all is in same area or country among the group.In order further to reduce the transmission quantity of updating message, selectivity is upgraded copy when upgrading, and namely only often accessed replica node is upgraded.For fear of producing the copy update inconsistency, all need in copy set, vote before each the renewal, poll just can be upgraded greater than a half of replica node sum in the group, otherwise can not.
Concrete handling process can be divided into following step:
1. Qun division
Fig. 2 is that CDN-P2P merges replica server layer division group schematic diagram in the content distributing network.Session between the replica server provides service by SP (SP:Service Provider).Arbitrarily selected service providing node SP in network, all nodes in the replica server layer can be to its direct visit.SP at first appoints the replica server node of K the mutual distance 〉=D that gets in the network node that serves as a mark, and builds the group centered by flag node.The address of all flag node all is stored among the SP.Be formed by connecting in tree-like mode between flag node, namely selected flag node is root, and the flag node near with its distance is its child node, and the flag node near with these child nodes is again its child node, sets up tree by that analogy.The non-marked node needs to the address that SP obtains flag node before adding the group, and the group of the flag node that chosen distance is nearest adds.The node physical location is close among the group, and is in same area as far as possible.Node number≤M of each group, and the average distance of arbitrary node and other node (distance referred to herein as time-delay)≤C.The node that does not satisfy the two condition can not add the group, sets up new group thereby can become new flag node.After division was finished, the replica server layer formed L group G
i, i=1,2 ..., L.Node adds and will be labeled node behind the group and distribute the node of the near node of a part and its distance and a part and its distance as its neighbor node.Flag node is the intermediate node of each group of replica server layer contact, so also claim intermediate node (MN:Middle Node), i group G
iFlag node abbreviate MN as
i, i=1,2 ..., L.
Because flag node is taken on important role in each group, be responsible for the information interaction between each group, causing for the inefficacy that prevents flag node between each group can't normally associate, node can periodically send detect-message whether be used for the check mark node effective among the group, and first node of finding that original flag node lost efficacy replaces it to become new flag node.
The uncertainty that when group's division, distributes owing to node, the member that the group who has is contained is less, the member that the group who has is contained is more, may exist some group only to comprise several nodes, load is very little, and adjacent group comprises more relatively node with it, and load is higher, safeguard that expense is heavier, caused the unbalanced of load.Simultaneously because the flag node among the group can and then cause the position as the flag node of group center that variation has taken place because problems such as inefficacy be upgraded, for the enforcement of load balancing of the present invention provides condition.
The present invention is that the mode of utilizing group node to shift realizes load balancing, reaches the optimization of content distributing network, is specially:
As certain a group G
nIn the member node quantity of node during less than predetermined value M, group G
nFlag node MN
nCan be to the more adjacent group's of member flag node, as G
mFlag node MN
m, ask for the own nearer node member of some distances.Concrete mode is as follows: flag node MN
nDetect adjacent group G
mIn node and the distance of self, as detected node and place group G
mFlag node MN
mDistance greater than with group G
nFlag node MN
nApart from the time, then meet the condition of asking for, flag node MN
nRespectively to group G
mDescribed in detected node and flag node MN
mSend the group node transfer request, if during accepting of receiving that above-mentioned node the sends response message that group node shifts, then described detected node is transferred to crowd G
nIn.
2. the foundation of responsibility replica node tree and copy set
With the example that is transmitted as of the updating message of a certain copy of content C, the responsibility replica node tree of description copy C and the foundation of copy set.When central server send be updating message at copy of content C the time, in each group, the node of content copy C is the replica node N of this copy of content C
c, otherwise be non-replica node.Wherein, first receives the replica node from the described updating message of central server or other group among the group, be called responsibility replica node or responsibility node (KN:Key Node), other replica node is called common replica node or ordinary node (ON:Ordinary Node).Therefore, i group G
iThe responsibility replica node of copy of content C abbreviate KN as
Ci, i=1,2 ..., L, i group G
iCommon replica node abbreviate ON as
Ci, i=1,2 ..., L.
(1) foundation of responsibility replica node tree: suppose that first receives the replica node N from the updating message of central server
c 1The group at place is G
1, this replica node N then
c 1Be responsibility replica node KN
C1KN
C1Can pass through its place group G
1Flag node MN
1Message is passed to the KN of other group, form with replica node N
c 1Responsibility replica node tree T for root node
c
For example, to receive the N of central server updating message
c 1For root is set up n fork responsibility replica node tree T
c, select n nearest KN for its child node, choose n again
2The near KN of individual distance is the child node of this n child node, sets up the tree that is made of KN by that analogy.Each KN contains his father, child node and root node address.
(2) foundation of copy set: when setting up responsibility replica node tree, the responsibility replica node KN of each group
CiCollect this group G at the appointed time
iIn contain the IP address of all replica node of copy of content C, set up copy set P
CiTherefore, each group G
iUnique copy set P corresponding to a copy of content C
Ci, i group G
iThe responsibility replica node be copy set P
CiThe responsibility replica node, i group G
iCommon replica node be copy set P
CiCommon replica node.Copy set P
CiInclude place group G in the information
iIn all replica node N
cAddress information, both comprised responsibility replica node KN
CiThe address, also comprise all ON
CiThe address.Concrete method for building up is: be loaded into the address information of oneself in the group Gi in the updating message of KNi with described copy content C, and send the updating message that is loaded with own address information, ON with the neighbor node pass-through mode
iThe feedback information that receiving gives out a contract for a project after this message contains own address is given the KN of this group
iIn order to set up copy set P
CiKN
iSet up copy set P in time at TTL
CiThe back is copy set P
CiInformation is issued this group ON
iFig. 3 is the procedure chart of setting up of the interior copy set of group.
Fig. 4 is the structure chart that builds among the group behind the copy set.
In order to alleviate the burden of MN, the KN in each group generally is chosen for interior first of group and receives the non-marked node of building group message.
Responsibility replica node, the responsibility replica node tree of corresponding different content copy C are different with copy set.The propagation that utilizes the updating message of copy of content C for the first time when the updating message of meaningful copy C arrives again, is set T in the responsibility replica node of this copy of content C correspondence earlier after responsibility replica node tree and copy set are set up
cIn begin successively to transmit updating message from root node, each KN in the responsibility replica node tree
CiBe broadcast to place group G after receiving again respectively
iMiddle copy set P
CiIn replica node.
3. safeguard copy set
Though the stability of replica server is very high, also may collapse and lose efficacy or because of the maintenance deviated from network, so copy set P
CiNeed to safeguard.Copy set P
CiMaintenance comprise that replica node adds, replica node is left with replica node and was lost efficacy.Wherein, add or deviated from network all needs to MN and the neighbor node message of sending out public notice at each node of replica server layer.And the adding of replica node with leave not only to notify above both, also to notify each replica node in the copy set, to avoid the transmission of unnecessary updating message.
(1) replica node is left: in the copy set replica node leave the two kinds of situations that comprise: leaving of responsibility replica node and leaving of common replica node.KN
CiNeed choose place copy set P before leaving
CiInterior strong common replica node of disposal ability is taken over and is become new KN
Ci, and then notice responsibility replica node tree T
cWith copy set P
CiIt leaves interior each common replica node; And ON
CiBefore leaving, only notify in the copy set other node its leave and get final product.
(2) replica node adds: replica node adds a certain copy set P
CiBefore, with the neighbor node pass-through mode to copy set P
CiAn interior replica node is sent out the message that joins request that contains own address, receives that the replica node in this copy set that adds message will contain copy set P
CiThe feedback message of information is issued replica node to be added, and notice copy set P
CiInterior all nodes have the newcomer to add message.
(3) replica node lost efficacy: lost efficacy for preventing replica node, at every turn at copy set P
CiBefore the interior transmission updating message, each responsibility replica node KN
CiAll need to place copy set P
CiInterior common replica node ON
CiSend and once ask updating message, receive common replica node ON
CiTo the response message of request updating message, no respondent is considered to it and leaves, is labeled as inefficacy by the responsibility replica node.Common replica node ON
CiPeriodic test place copy set P
CiResponsibility replica node KN
Ci, if responsibility replica node KN
CiLost efficacy, first finds the common replica node ON that it leaves
CiBe responsibility replica node KN
CiFor example, ON
CiPeriodically to KN
CiSend detect-message, if KN
CiWhen not returning the response message of detect-message, then think KN
CiLost efficacy, first assert KN
CiThe ON that lost efficacy
CiBe updated to place copy set P
CiNew responsibility replica node KN
CiNew KN
CiPass through MN
iThereby the KN that contacts other group adds responsibility replica node tree T
c, finish the renewal of responsibility replica node tree.
4. the improvement of copy set
Further, in order to reduce the transmission of updating message, the present invention also proposes the method to replica node selective updating in the copy set: each copy set P
CiIn a visit frequency threshold value is set, copy set P
CiEach interior replica node can check the access frequency of oneself when each updating message arrives, continue to be retained in the copy set when access frequency is higher than renewal frequency; When finding that own access frequency is less than or equal to renewal frequency, replica node can be left copy set automatically, and the copy content C of mark oneself is invalid and notify replica node in the copy set with this copy content C invalid message.Its status is the same with non-replica node.When being higher than renewal frequency, the replica node access frequency of copy content invalid can add copy set P again
Ci
5. update inconsistency avoids
In order to prevent the appearance of update inconsistency, adopted voting mechanism when the renewal of the present invention copy of content C in carrying out copy set and visit.
At responsibility replica node KN
CiThe updating message broadcast transmission is arrived place copy set P
CiCommon replica node before, responsibility replica node KN
CiCan be to copy set P
CiIn all replica node initiate update requests, receive that the replica node of update request responds, if copy set P
CiIn replica node number of responses (also claim write votes) WQ less than 1/2 of N, temporarily do not send updating message, up to WQ〉WQ node broadcasts to response during 1/2*N send updating message, the node that does not have response is considered as leaving copy set, wherein, N is crowd G
iIn all comprise the replica node number of effective copy of content C, namely comprise replica node (being online replica node) in the copy set, leave the replica node of copy set and the replica node of inefficacy, and do not comprise that the copy content-label is invalid replica node.This external statistics replica node number of responses WQ(writes votes) time, also can be with responsibility replica node KN
CiCount.
As the above-mentioned copy set P of other replica server (the non-replica node that for example, does not contain meaningful C in copy set place group or other groups) visit
CiIn copy of content the time, in order to guarantee that the copy of content that other replica server obtains is latest edition, also need to copy set P before reading
CiAll interior replica node are initiated the request of reading, if replica node number of responses (also read votes) RQ〉N-WQ, having the data of a replica node at least in the node that then guarantees to respond is latest edition, selects the highest the getting final product of version in the replica node data when replica server obtains content; If do not satisfy RQ〉N-WQ, then can not read, the copy set of a group does not satisfy condition and can read to other group's who satisfies condition copy set yet when reading data.
If do not set above-mentioned condition, the update inconsistency of copy of content appears easily.For example, when original 7 replica node A, B, C, D, E, F, G were arranged in the copy set, 7 replica node all had the prototype version of copy of content.When carrying out the renewal of version 1, only have that 3 replica node A, B, C are online to write ballot; When carrying out the renewal of version 2, only have that 3 replica node D, E, F are online to write ballot; The copy that replica node A, B, C have after then upgrading is version 1, and the copy that replica node D, E, F have is version 2, and the copy that replica node G has is prototype version, and each node can think that all own copy of content version is new.When if other non-replica node desires are visited the copy content of this copy set, if having not online the reading of D, E, the F replica node of latest edition 2 votes, only can access node A, B, the copy of content of C or G, but not replica node does not know also to exist the copy of content of version 2, therefore can't learn that the copy of content of node A, the B, C or the G that visit is out-of-date.And when defining RQ〉behind the access consideration of N-WQ, can guarantee to respond desire and visit in RQ the replica node in the copy set of non-replica node and have a latest edition with copy of content at least, desire is visited non-replica node node can read the highest copy of content of version from this RQ replica node.In addition, when WQ being defined as greater than N/2, RQ then value is less relatively, is conducive to reach fast the RQ value, realizes desiring the access to content of access node.
In the present invention, flag node is the Centroid of each group, is serving as important role, therefore needs the average distance node near and that disposal ability is strong of each node among selection and the group to serve as.Node periodic test flag node among the group is if there is the alternative flag node of the node that is more suitable for.In like manner, in copy set, the responsibility replica node is responsible for updating message is passed to replica node among the group, also needs the strong node of disposal ability to serve as.The common replica node that disposal ability is strong also can " be recommended oneself " and be served as new responsibility replica node.This method is applicable to the network that node stability is strong, and the server stability of replica server layer is all very high, therefore need not too many maintenance costs when safeguarding.
The present invention in identical group, and sets up copy set with physical location is close in the replica server layer node division in each group.Replica node in copy set that access frequency is low shifts out copy set, has further reduced message propagation quantity.When updating message arrives, begin successively to transmit updating message in responsibility replica node tree from root node earlier, after receiving, each responsibility replica node is broadcast to the node in each copy set more respectively.Taked " ballot " mode for preventing the copy update inconsistency.The method can effectively reduce the transmission delay of updating message, and redundancy message quantity also can significantly reduce, and has also avoided the generation of update inconsistency and has guaranteed that other replica server can obtain effective content simultaneously.
Claims (9)
1. a CDN-P2P merges the copy of content C consistency maintaining method based on physical location of content distributing network, described CDN-P2P merges content distributing network and comprises central server layer, replica server layer and client layer, it is characterized in that, may further comprise the steps:
Step 1, merge replica server layer in the content distributing network at CDN-P2P, optional a plurality of nodes node that serves as a mark, and centered by each flag node, set up a plurality of crowds of G
i, i=1,2 ..., L, L are group's sum, wherein pass through flag node MN separately between each group
iCommunicate;
Step 2, first receives the replica node N at the updating message of copy of content C from central server
c 1Described updating message is passed to the replica node N of other group by flag node
c, wherein, replica node N
cBe the node of content copy C, first replica node that receives described updating message is place group G in each group
iResponsibility replica node KN
Ci, i=1,2 ..., L; Travel through the responsibility replica node KN of each group
CiFoundation is with replica node N
c 1Responsibility replica node tree T for the copy of content C of root node
c, wherein, each responsibility replica node has its father node, child node and root node address;
The responsibility replica node KN of step 3, each group
CiObtain place group G
iThe replica node N of middle content copy C
cThe copy set P of copy of content C is set up in the address
Ci, i=1,2 ..., L, and with described copy set P
CiInformation sends to copy set P
CiEach interior common replica node ON
Ci, i=1,2 ..., L, wherein, common replica node ON
CiBe copy set P
CiIn except responsibility replica node KN
CiOther outer replica node;
Step 4, the utilization propagation of the updating message of copy of content C are for the first time set T with the responsibility replica node
cWith copy set P
CiAfter setting up well, when the updating message of meaningful copy C arrives again, from the responsibility replica node tree T of copy of content C
cRoot node N
c 1Beginning is successively transmitted, responsibility replica node tree T
cIn each responsibility replica node KN
CiReceive the copy set P that is broadcast to the place after the described updating message more respectively
CiIn.
2. a maintaining method as claimed in claim 1 is characterized in that, responsibility replica node KN
CiBe group G
iIn first receives the non-marked node of described updating message.
3. a maintaining method as claimed in claim 1 is characterized in that, the mode of setting up a plurality of groups centered by flag node is: the group of the flag node that non-marked node chosen distance is nearest adds; The node number is less than threshold value M among each group, and in arbitrary node and the group average distance of other node less than threshold value C.
4. a maintaining method as claimed in claim 1 is characterized in that, group G
iIn each node periodic test place group's flag node MN
iWhether effective, first determines this group mark node M N
iInvalid node updates is new flag node.
5. a maintaining method as claimed in claim 1 is characterized in that, also comprises copy set P after the step 4
CiIn replica node leave safeguard step: when the replica node of leaving is responsibility replica node KN
CiThe time, responsibility replica node KN
CiSpecify new responsibility replica node in advance, and to responsibility replica node tree T
cWith copy set P
CiInterior each common replica node ON
CiSend leave group message; Be common replica node ON when leaving replica node
CiThe time, common replica node ON
CiIn advance to copy set P
CiInterior other replica node sends leave group message.
6. a maintaining method as claimed in claim 1 is characterized in that, also comprises copy set P after the step 4
CiIn replica node add safeguard step: the node that desire adds is to copy set P
CiAn interior replica node sends the message that joins request, and receives that the replica node of the described message that joins request contains place copy set P to the node transmission that described desire adds
CiThe feedback message of information, and to copy set P
CiInterior other replica node sends newcomer and adds message.
7. a maintaining method as claimed in claim 1 is characterized in that, also comprises copy set P after the step 4
CiIn replica node lost efficacy safeguard step: at each copy set P
CiBefore the described updating message of middle broadcasting, responsibility replica node KN
CiTo place copy set P
CiIn each common replica node ON
CiSend the request updating message, receive each common replica node ON
CiResponse message, and will not have the common replica node ON of response message
CiBe labeled as inefficacy;
Common replica node ON
CiPeriodic test place copy set P
CiResponsibility replica node KN
Ci, if responsibility replica node KN
CiLost efficacy, first finds the common replica node ON that it leaves
CiBe new responsibility replica node, the responsibility replica node that new responsibility replica node contacts other copy set adds responsibility replica node tree T again
c
8. a maintaining method as claimed in claim 1 is characterized in that, copy set P is set
CiThe node visit frequency threshold, as copy set P
CiWhen the node visit frequency of interior replica node was lower than described threshold value, described replica node was from copy set P
CiInterior deletion, and the copy of content C of the described replica node of mark is invalid.
9. a maintaining method as claimed in claim 1 is characterized in that, at each copy set P
CiBefore the interior broadcasting updating message, responsibility replica node KN
CiTo place copy set P
CiIn all nodes send request updating message, as request updating message number of responses WQ〉during 1/2*N, WQ node broadcasts of response sent updating message, wherein N is crowd G
iIn the effective replica node number of copy of content;
As non-replica node visit copy set P
CiIn copy of content the time, described non-replica node is to copy set P
CiIn replica node initiate to read request message, when reading request message number of responses RQ〉during N-WQ, described non-replica node reads the highest copy of content C of version from RQ replica node of response.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310137231.2A CN103200273B (en) | 2013-04-19 | 2013-04-19 | The copy of content consistency maintaining method of physically based deformation position |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310137231.2A CN103200273B (en) | 2013-04-19 | 2013-04-19 | The copy of content consistency maintaining method of physically based deformation position |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103200273A true CN103200273A (en) | 2013-07-10 |
CN103200273B CN103200273B (en) | 2016-01-20 |
Family
ID=48722628
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310137231.2A Expired - Fee Related CN103200273B (en) | 2013-04-19 | 2013-04-19 | The copy of content consistency maintaining method of physically based deformation position |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103200273B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103560897A (en) * | 2013-11-05 | 2014-02-05 | 腾讯科技(武汉)有限公司 | Overall broadcasting method, server and system |
CN105897456A (en) * | 2015-12-07 | 2016-08-24 | 乐视云计算有限公司 | Content distributing method and system |
CN107045426A (en) * | 2017-04-14 | 2017-08-15 | 北京粉笔蓝天科技有限公司 | A kind of many copy read methods and system |
CN108234173A (en) * | 2016-12-21 | 2018-06-29 | 中国电信股份有限公司 | Node Replica updating method and apparatus |
CN109981777A (en) * | 2019-03-27 | 2019-07-05 | 深圳市网心科技有限公司 | Copy maintaining method, server, copy memory node and system based on CDN |
CN110188006A (en) * | 2019-06-03 | 2019-08-30 | 深圳市网心科技有限公司 | A kind of data maintaining method, device and equipment |
CN111064780A (en) * | 2017-03-01 | 2020-04-24 | 贵州白山云科技股份有限公司 | Multitask content updating method, device, equipment and medium |
CN111696245A (en) * | 2020-06-30 | 2020-09-22 | 郭平波 | Voting method based on P2P network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101841531A (en) * | 2010-03-16 | 2010-09-22 | 中国科学院计算技术研究所 | Simulating method and system for CDN-P2P (Content Distribution Network-Peer-to-Peer) hybrid network |
CN101841556A (en) * | 2010-02-23 | 2010-09-22 | 中国科学院计算技术研究所 | Method and system for placing resources replication in CDN-P2P (Content Distribution Network-Peer-to-Peer) network |
CN102065150A (en) * | 2011-01-18 | 2011-05-18 | 乐视网信息技术(北京)股份有限公司 | Data transmission system and method based on P2P (Peer-to-Peer) network and CDN (Content Delivery Network) |
CN103023928A (en) * | 2013-01-11 | 2013-04-03 | 乐视网信息技术(北京)股份有限公司 | P2P (peer-to-peer) node matching system and method |
-
2013
- 2013-04-19 CN CN201310137231.2A patent/CN103200273B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101841556A (en) * | 2010-02-23 | 2010-09-22 | 中国科学院计算技术研究所 | Method and system for placing resources replication in CDN-P2P (Content Distribution Network-Peer-to-Peer) network |
CN101841531A (en) * | 2010-03-16 | 2010-09-22 | 中国科学院计算技术研究所 | Simulating method and system for CDN-P2P (Content Distribution Network-Peer-to-Peer) hybrid network |
CN102065150A (en) * | 2011-01-18 | 2011-05-18 | 乐视网信息技术(北京)股份有限公司 | Data transmission system and method based on P2P (Peer-to-Peer) network and CDN (Content Delivery Network) |
CN103023928A (en) * | 2013-01-11 | 2013-04-03 | 乐视网信息技术(北京)股份有限公司 | P2P (peer-to-peer) node matching system and method |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015067151A1 (en) * | 2013-11-05 | 2015-05-14 | 腾讯科技(深圳)有限公司 | Global broadcast method, server and system |
CN103560897B (en) * | 2013-11-05 | 2016-07-27 | 腾讯科技(武汉)有限公司 | A kind of overall situation broadcasting method, server, and system |
CN103560897A (en) * | 2013-11-05 | 2014-02-05 | 腾讯科技(武汉)有限公司 | Overall broadcasting method, server and system |
CN105897456A (en) * | 2015-12-07 | 2016-08-24 | 乐视云计算有限公司 | Content distributing method and system |
CN108234173A (en) * | 2016-12-21 | 2018-06-29 | 中国电信股份有限公司 | Node Replica updating method and apparatus |
CN108234173B (en) * | 2016-12-21 | 2021-05-04 | 中国电信股份有限公司 | Node copy updating method and device |
CN111064780B (en) * | 2017-03-01 | 2023-04-18 | 贵州白山云科技股份有限公司 | Multitask content updating method, device, equipment and medium |
CN111064780A (en) * | 2017-03-01 | 2020-04-24 | 贵州白山云科技股份有限公司 | Multitask content updating method, device, equipment and medium |
CN107045426B (en) * | 2017-04-14 | 2020-12-15 | 北京粉笔蓝天科技有限公司 | Multi-copy reading method and system |
CN107045426A (en) * | 2017-04-14 | 2017-08-15 | 北京粉笔蓝天科技有限公司 | A kind of many copy read methods and system |
CN109981777A (en) * | 2019-03-27 | 2019-07-05 | 深圳市网心科技有限公司 | Copy maintaining method, server, copy memory node and system based on CDN |
CN110188006B (en) * | 2019-06-03 | 2022-04-29 | 深圳市网心科技有限公司 | Data maintenance method, device and equipment |
CN110188006A (en) * | 2019-06-03 | 2019-08-30 | 深圳市网心科技有限公司 | A kind of data maintaining method, device and equipment |
CN111696245A (en) * | 2020-06-30 | 2020-09-22 | 郭平波 | Voting method based on P2P network |
Also Published As
Publication number | Publication date |
---|---|
CN103200273B (en) | 2016-01-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103200273B (en) | The copy of content consistency maintaining method of physically based deformation position | |
KR101544480B1 (en) | Distribution storage system having plural proxy servers, distributive management method thereof, and computer-readable recording medium | |
CN100558042C (en) | A kind of P2P live broadcasting method based on super node | |
CN101133622B (en) | Splitting a workload of a node | |
CN101938508B (en) | Method and system for shortening time delay in peer-to-peer network streaming media live broadcast system | |
CN101567796B (en) | Multimedia network with fragmented content and business method thereof | |
CN100588172C (en) | System and method for realizing network reserved storage | |
JP4473942B2 (en) | Content distribution apparatus, content distribution method, and content distribution program | |
CN101610162A (en) | A kind of method, system and equipment that content is provided based on reciprocity storage networking | |
CN103095727B (en) | P2p resource location method | |
KR101485610B1 (en) | Distributed content delivery system based on network awareness and method thereof | |
CN102843420A (en) | Fuzzy division based social network data distribution system | |
CN102523256A (en) | Content management method, device and system | |
CN102316097B (en) | Streaming media scheduling and distribution method capable of reducing wait time of user | |
CN101072160B (en) | Distributed virtual environment management method, system and node | |
CN104539982A (en) | Method and system for dispatching of video resource point-to-point sharing and node terminal | |
JPWO2008120366A1 (en) | Content distribution apparatus, content distribution method, and content distribution program | |
CN101931638A (en) | Method for building service for recommending data content to Internet user | |
CN100473007C (en) | Content exchange network | |
CN101217565A (en) | A network organization method of classification retrieval in peer-to-peer network video sharing system | |
CN101854387A (en) | P2P flow optimization method and system based on distributed index server architecture | |
Ebadi et al. | A new distributed and hierarchical mechanism for service discovery in a grid environment | |
CN103179191B (en) | P2P network control device and P2P network managing and control system | |
CN100559758C (en) | Method based on building combination P2P system | |
CN101668037A (en) | Method for dispatching P2P network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160120 Termination date: 20190419 |