CN101645872B

CN101645872B - Business service system of Internet large-scale application environment and working method thereof

Info

Publication number: CN101645872B
Application number: CN 200810117823
Authority: CN
Inventors: 李冰
Original assignee: Peking University
Current assignee: Peking University
Priority date: 2008-08-05
Filing date: 2008-08-05
Publication date: 2012-12-19
Anticipated expiration: 2028-08-05
Also published as: CN101645872A

Abstract

The invention discloses a business service system of Internet large-scale application environment and a working method thereof and belongs to the field of distributed technology. The system comprises a server and a client, wherein the server is provided with a business logic interpreter, a business logic and business data; the client is provided with a business logic interpreter; and the server and the client are connected by a network. The method comprises that: firstly, the server receives a business request of the client and transmits the business logic and business data to the client according to the request; and secondly, the server establishes a dynamic cluster according to the business service type of the client and performs the state maintenance and business transaction management of the dynamic cluster through multicast. The method reduces investment in Internet-based business server computing resources and realizes large-scale business computing at the same time. The system is low in maintenance cost and has personalized characteristics.

Description

The commercial service system and the method for work thereof of a kind of the Internet large-scale application environment

Technical field

The present invention relates to a kind of commercial service system and method for work thereof; Especially pay close attention to the commercial service system and the method for work thereof of a kind of the Internet large-scale application environment; Belong to the application of distributed computing technology on enterprise software, its field that relates generally to comprises commercial server, multicasting technology, database technology and Clustering etc.

Background technology

The application of computing technique in the middle of enterprise is very general.The commercial server that the present invention relates to not is any server system that is used for enterprise application of general reference, is calculated as the commercial server of characteristic based on Internet technology but refer in particular in the The whole calculations process with lightweight data and business logic.The major technique and the relevant issues of current commercial server are following.

(1) all operations under the extreme confined condition of computational resource of current commercial server.Enterprise application system is different from nonbusiness's application and is that mainly enterprise application requires to be absolutely correct and safety.Under such demand, system would rather any sacrifice in performance will guarantee also that business process do not make mistakes.The Traditional business application environment generally carries out in enterprises; Computational resource and relevant participant are under strict control like this, and it is just simple relatively to achieve the above object.Therefore, system can have more design to select in such environment, needn't too strictly be limited by problem of management.Such as, server can bear too much calculating pressure, can on client, arrange some calculating; Even also can cooperate each other between the client and form than the more effective topological structure of client server.After Internet technology was popularized gradually, enterprise application turned to gradually with the Internet and combines; The environment of commercial application just changes opening into by relative closure like this.In this open environment, the maintenance of correctness and fail safe just need be paid higher cost; If handle according to the mode of enterprises fully, above-mentioned two principles just can't guarantee.Therefore, current commercial server has taked all calculating all on server, to carry out, and client only is responsible for providing in the calculating necessary parameter not participate in any Practical Calculation.This mode but has to server is depended on unduly to guaranteeing the correct favourable with safety of business system.At this time server will carry all commercial pressure that calculates.When the client scale enlarged rapidly, the resource of server end can be more and more limited.For this reason, server end design cost can significantly improve, and has to take complex scheduling algorithm to come effectively to use limited computational resource.When client access peaks, ultimate attainment even dispatching algorithm is accomplished, also because the deficiency of computational resource causes period of reservation of number long, even system crash.

(2) current commercial server cost performance is low.In order to support potential a large number of users, must guarantee the abundance of computational resource based on the commercial systems for applications of the Internet with the mode of high investment.The mode of generally taking comprises rents high bandwidth, purchases high-performance server, adopts complicated Clustering and arranges buffer memory equipment or the like.In addition, the design of software corresponding technology also needs high investment, under limited computational resource condition, guarantees user capture as much as possible as far as possible.Such business system is the great amount of calculation resource overhead obtaining high performance behind, has formed the low excessively reality of cost performance.This phenomenon is to compare with the system of tradition based on enterprises flexible design structure; Certainly, compare with solution of the present invention, current commercial server cost performance is also much lower.

(3) current commercial server exists computational resource serious waste phenomenon.The main problem of current commercial system embodies a concentrated reflection of on the commercial server owing to its particular structural.But when computational resource was limited, also there was the contradictory phenomena of computing resource waste in current commercial server.This mainly is to be the pressure that under high access or even the extreme case server is caused because current commercial server considers in when design.This situation occurrence probability most of the time is not high.This has caused a large amount of inputs only for tackling the drawback of small probability event.And usually the generation of extreme case is difficult for estimating, and deals with when really high access occurring and has any problem.The way of generally taking is to adapt to the average access amount; In case when high access occurring, increasing computational resource or adjustment algorithm relevant parameter.When visit capacity increases when rapid, this process can not guarantee dynamically adapting; User capture has to postpone even lost efficacy at this moment.Current commerce server is calculated in the system scale variation in adaptation and is not reached strong retractility.This also be the present invention one of the problem that will try hard to solve.

(4) a little less than the personalized service ability based on current commercial server.As the business system that has a large number of users and mass data, be necessary for the user and provide personalized service, make it from bulky systems, to find as early as possible the commerce services that satisfies its sexual demand.Yet, since to user collective demand can only through to before user's historical record analyze, make that the foundation of accomplishing this target is insufficient; Secondly, this analysis is again to be based upon on the jejune machine algorithm, causes that accuracy rate is low as a result; When a large number of users all with different demands simultaneously during login system, system has to pay higher computational resource for this reason, the system that makes is difficult to bear.

(5) current commercial server management advantage is also disappearing gradually.For the large scale business application system, its server scale enlarges gradually, and causing managing cost increases, like consistency maintenance, fault-tolerance maintenance, performance assurance, fail safe, state and transaction management or the like.When system scale was little, this management cost wanted much little; After system scale enlarged, numerous computing nodes were distributed on many places.Management role has become distributed management by unified management gradually.In addition; Raising along with computing equipment, the communication technology and related software technology; Network between commerce server and the client connects and is more and more convenient alternately, the computing capability of client also from strength to strength, the application system that is based upon on these equipment is tending towards relatively stable.If possible calculate in the enterprising industry of doing business of such system, its cost was much smaller than the past.This also is the place that the present invention paid close attention to.

(6) current commercial server also faces the pressure of heavyweight data.Most of business systems all are to be applied as the master with lightweight.But along with the expansion of system scale and the user raising to quality of service requirement, business system also is faced with the pressure that the heavyweight data are used.This pressure is on the one hand from business data itself, on the other hand also from the commerce services based on media data.The calculating pressure that produces therefrom makes that more business system can't bear the heavy load.

Summary of the invention

The objective of the invention is to propose a current commerce server of reduction drops into relatively, significantly improves business computing system on the guaranteed performance simultaneously.This new system will possess the ability that strong retractility, high performance-price ratio, personalized service, management are convenient relatively and handle the heavyweight data.Except being the large scale business system saves input, can also cheap commercial computing environment be provided for medium-sized and small enterprises.

The commercial service system of a kind of the Internet large-scale application environment comprises server end and client, it is characterized in that said server end deployment business logic interpreter, business logic and business data; Said client deployment has the business logic interpreter; Said server end is connected through network with said client.

The commercial service system works method of a kind of the Internet large-scale application environment the steps include:

1) adopt TCP/IP peer protocol pattern to make computing node have the ability of the request of reception and transmission request;

2) server end receives the commerce request of client and moves to this client to business logic;

3) server end is according to the request transmission business data of this client;

4) server end is set up dynamic cluster according to the commerce services classification of client; Said dynamic cluster is one group of client with common business data;

5) in above-mentioned dynamic cluster, carry out state-maintenance and business management through multicast.

Adopt the SSL method that said business logic is moved to client in the said method.

The transmission method of said transmission business data is: adopt the method for buffer memory progressively business data to be moved to client from server end.

Set up said dynamic cluster according to the drift net agreement in the said method; Adopt the Pareto optimality principle to carry out the cooperation between the node in the said dynamic cluster.

Multicasting method in the said dynamic cluster is that its method is based on the lightweight multicast that improves the BT agreement:

1) uses tracker monitoring dynamic cluster node state; Said state comprises whether node is in the system, node whether to certain data pay close attention to, whether node be whether super node, node have obtained these data;

2) tracker distributes the multicast task according to detecting the state equalization that obtains, and its method is:

A) tracker confirms that according to the formed historical information of node line duration length the node of right quantity is a super node;

B) let these super nodes at first on the lenient side the magnitude data produce node and obtain data;

C) said super node is on average shared the multicast task of residue node and is returned to tracker to correlation behavior information.

In the said method, tracker is coordinated said super node and is carried out the multicast task, and its method is:

1) when a super node is attempted to send data, at first to send request to tracker;

2) tracker sorts according to the request sequencing to all nodes that need send data in the cluster;

3) tracker carries out asynchronous transmission then to being carried out serial number by sorting data;

When 4) each node was handled the data that obtain, whether judgment data lost, if find that numbering is discontinuous, then got in touch with data source and obtained related data.

In the said method, said business management comprises:

1) the server end concurrent operations that possibly clash is changed into the serial operation that meets sequential, and its method is: at first node is with the business executing state announcement server end that self comprises write operation; Server end is added up the transaction write mode of operation then, is registered by the affairs of write data with having jointly chronologically; Last server end is responded interdependent node and whether can be carried out write operation at once or postpone the products for further notice according to transaction management SECO principle;

2) accomplish the recovery after affairs are abandoned through the control of transaction write time sequential routine and to the monitoring that node logs off; Its method is: the node that carries out the affairs registration at server end periodically sends heartbeat data to its neighbours; If its neighbours do not receive heartbeat data in the fixed cycle; This can think that this node is improper and withdraws from, and promptly abandons affairs; Afterwards, these neighbours understand the proactive notification server end to accomplish the data necessary recovery; If node normally withdraws from, server end is normally carried out recovery operation.

In the said method, when new user access server, server forwards the data to the associated dynamic cluster more in view of the above according to the commerce services that the user asks judges to be concerned about, makes it obtain related service.

Said node comprises the client and server end.

Basic thought of the present invention is to be based upon to utilize all possible computational resource to accomplish commercial calculating.It is the description of its business logic that current commercial server provides the core of commerce services.This business logic all is to concentrate to be positioned on the commercial server in current system: no matter being to explain to carry out or the compiling execution, all is under the control of commerce server.This business logic under the commercial server centralized management can only use its host's computational resource to respond user's request.The execution of business logic needs the consumption calculations resource and a time of implementation is arranged.In order to improve user's request responding speed, commercial server generally must adopt asynchronous or parallel mode.When parallel the generation, the management cost can improve, also can be higher to the requirement of computational resource.When the increase of user capture quantity was too rapid, server was had to transfer more resources and is accomplished concurrent operation; Its management cost also can further increase.Under the extreme case, server resource is unable to cope with too much request, and period of reservation of number will infinitely prolong, even causes server resource to exhaust and collapse.Can find out from above discussion, if can the dispersion user request to different computing equipments, just can slow down resource pressure to server.The dispersion user request mode is through arranging a plurality of servers and renting enough bandwidth and realize in current commercial the calculating.For business logic, also must be arranged on these servers.This process is artificial the completion fully.Above solution be that to drop into more resources with the ISP be that characteristic strengthens commercial computing capability.The present invention has taked other a kind of mode, promptly gives business logic transportable ability, thereby reaches the purpose that avoids manual work and finish the work and shift in running status; Further, the present invention does not need the additional arrangement server, but is arranged into transportable business logic automatically on the computing equipment of user's control.So just avoided input pressure to the commerce services supplier.

The first step selects interpretable programs as the migration scheme.Realize the migration of business logic, two kinds of schemes can be arranged.First kind of scheme is the Direct Transfer executive program, the object code that promptly obtains after the compiling; Second kind of scheme is that the business logic that only moves soluble operation is described.The two can accomplish the migration to business logic.

(1) moves the problem that object code need be considered.The first, the object code size will be much larger than interpretable programs, and the network cost is big when causing transmission.Secondly, object code requires absolute and former host compatible mutually to the migration host.This requires in large-scale commercial applications is used, possibly can't satisfy; A large scale business system generally includes numerous heterogeneous clients, and this can cause object code on some clients, not move.Once more, for some business logics that needs dynamically adapting, be that former host or new host all can't be changed object code; This can cause can't normally moving of business logic.At last, object code is wayward.Even object code is migrated to the client of complete compatibility, but because object code is in the isolated operation state, but with the client original system when mutual selection mode limited; Can't intervene for object code consumption of natural resource and correlation behavior management.This takes flexible algorithm management own resource to cause difficulty to client.The advantage of migration object code is that the fail safe of business logic is guaranteed easily, and operational efficiency is high.In addition, with respect to interpretable programs, its characteristic of carrying out separately also can be simplified the design of client original system.

(2) move the problem that interpretable programs need be considered.Compare with the migration object code, the migration of interpretable programs possesses following advantage.The first, the migration of interpretable programs is little to network pressure, and this is that its limited size belongs to the lightweight data because interpretable programs is the description to business logic.The second, interpretable programs can be moved on any host that possesses the relevant explanation device, for large-scale commercial applications is used, can form the effect that business logic and underlying platform have nothing to do.The 3rd, after interpretable programs is migrated to client, in some commercial application, can be used to on-the-fly modify to adapt to the concrete variation in the business process.The 4th, interpretable programs is meeting and original system seamless combination after being moved, and becomes the part of original system; Host machine system can be taked and the same strategy of own resource is managed.Yet the migration of interpretable programs also has following problem.The first, operational efficiency is low.This must explain owing to its logic and carry out simultaneously and cause; This interpretation process needs consumption of natural resource and time.The second, fail safe is slightly poor.No matter interpretable programs is in the middle of the migration on the network, or when on host, moving, all can be because of itself being the leakage that the description of business logic is caused business secret.The 3rd, the cost of interpreter is big.Though interpretable programs can accomplish and platform independence that its prerequisite is on any potential client or host, interpreter to be installed.This cost wants big than executable program.

(3) to the selection of migrator.In sum, be that object code or interpretable programs all have Pros and Cons separately.Because the object code migration has the difficulty that can't overcome, as moving in some clients, this is unacceptable for the large scale business system, takes to move interpretable programs and is inevitable.Simultaneously, the common complexity of commercial calculating is not high, even the interpretable programs operation is slow relatively, and its relative efficiency or acceptable.The present invention need transform the client and server end, in this is transformed, embeds relative commercial logical interpretation device and can not let the user pay added burden.On safety issue, this is the problem that the distributed business system all can face, and can take (the reference: SSL 3.0Specification of SSL method; Http:// wp.netscape.com/eng/ssl3/) safe transmission of assurance business logic.In the difficulty that has solved the migration interpretable programs and after combining the advantage of interpretable programs itself, it is exactly rational selecting interpretable programs to move.

Second goes on foot, and sets up the corresponding relation of data and business logic in client through the mode of complete migration business logic and buffer memory business data.In commercial computing environment, any one business logic all is to act on related data.When migration took place business logic, related data also must be moved thereupon, otherwise can occurrence logic and the phenomenon of the different physical locations in data office; Because network delay can make systematic function descend.This requires the task that commerce is calculated is divided, and is mapped different business logics and data respectively, can accomplish the low coupling of high cohesion between the different task as far as possible.This is to developer's requirement on stream.But a system that possesses business logic and data transfer ability can not too rely on developer's work.Development itself is unusual complicated, tries not need bring added burden to exploitation because of what move.In practical application, can take an acceptable scheme to get final product usually.At first this transmission cost in migration of business logic is low, even therefore complete transmission can not cause system too much to bear yet.Secondly,, can take the method for buffer memory for related data, progressively data from the server migration to the client, rather than once all move related data.This way begins to have repeatedly between meeting and the server mutual, even can influence client end response speed.Along with related data is backuped to client, corresponding speed can improve gradually but afterwards; The more important thing is that the pressure of migration back server reduces greatly.

In the 3rd step, be based upon the dynamic cluster under the commercial computing environment in the Internet.The foundation of dynamic cluster is one of key of the present invention.Set up dynamic cluster and calculate the basis, at first need make each node when having request msg and service ability, also possess the ability of other node requests of response.Achieve this end not difficultly, utilize basic network technology [1] just can accomplish.Difficulty is to break through the NAT obstacle that exists on the Internet, and its method can be with reference to [2].The solution of this problem can be so that can be directly mutual each other between any two nodes in the Internet.This is to produce the prerequisite that large-scale internet calculates.Secondly, such node also need improve its throughput except can asking and respond, and promptly under the enabled condition of node computational resource, can establish a communications link with a plurality of nodes concomitantly, and can in such connection, continue the transmission data.The important technology here is exactly the realization [3] of thread pool.At last, state target in realization after, also having an important process is exactly to give node to seek the ability [4] of cooperative nodes.In a large-scale internet system, any one node can only be connected and cooperate calculating with limited node.Should be that a node searches out optimal node in numerous on the internet computing nodes, thereby realize that computational resource utilizes to greatest extent.Dynamic cluster refers to one under running status, and based on common calculating purpose, the computational resource of formation is shared set by various on the internet the computing equipments of possibly connecting.This set has following characteristic.

(1) life cycle.Dynamic cluster has life cycle.Common dynamic cluster all to pass through generation, development, maturation, steadily, desolate and decay a plurality of stages.But not all dynamic cluster all will pass through these complete processes.Or some cluster directly decay crossing over some stage after the generation can be in generation for a long time and decay outside certain stage.The generation of dynamic cluster is to initiate a calculating through a computing node, causes the response of other nodes, thereby forms cluster; When responsive node increased, cluster just was in developing stage; Developing stage can arrive a limit, and Here it is the stage of ripeness; After the maturity period, cluster scale can tend to balance gradually, and scale no longer enlarges, and does not also obviously dwindle, i.e. plateau; Along with most nodes are accomplished calculating and left this cluster, the node in the cluster reduces gradually, thereby causes the formation in desolate stage; After all computing nodes all left in the cluster, cluster was withered away, and got into and decayed the stage.

(2) common purpose.The basis that dynamic cluster forms and keeps is common calculation task.Calculation task refers to the transmission that takes place in data or the service process, the mutual and computing activity of providing.A dynamic cluster comprises a plurality of nodes; These nodes all carry out correlation computations round one or more correlation computations tasks.They are the beneficiary of related data and service, also are the indirect suppliers of related data and service.

(3) independent assortment.Node in the dynamic cluster has the characteristic of independent assortment.No matter how the cluster scale size changes, and the node in the cluster all adds under to the interested condition of correlation computations task.After node has been enjoyed related data or service, perhaps lose the data or service that correlation computations provides after, then can leave system at any time and not receive any restriction.

(4) highly dynamic.The principal character of dynamic cluster is exactly that this cluster always is in the middle of the continuous variation.This variation both had been embodied in cluster itself and had had life cycle, was also embodied in node and constantly added or leave, even be embodied in the variation on the topological structure that cluster has.These behavioral characteristics reflect cluster state at once; Assurance to these states can be played important support effect for the system that builds on such cluster.

(5) mesh topology.As a kind of application form of distributed computing technology, any cluster all can show with a kind of form of topological structure.To the understanding of topological structure, also can hold appropriately to the characteristic of cluster.Dynamic cluster among the present invention shows as network structure.But this description is too rough.Understanding to this structure can be with reference to related article [5].Why forming network structure but not other topological structures, is because such structure can be so that can share computational resource separately between each node to greatest extent, thereby under the dynamic situation of height, realizes the maximization of calculated performance.Fig. 1 has represented the topological structure at once of dynamic cluster.

(6) highly heterogeneous.Though each node is based on common task and is organized into cluster, still has many differences in this cluster.What is called is highly heterogeneous to be showed by following characteristics.The first, the heterogeneity of node resource.Usually show between each node have the difference of computational resource; And the node that has a various computing resource is different to the potential contribution of system.Should take appropriate algorithm [4,6] both to guarantee the normal operation of whole cluster, also can make each node can access the service on himself resource base of maximum using.The second, the heterogeneity of data or service.Data are to maintain the dynamic cluster life cycle to be in the basic of stable state with service.In the middle of High-Performance Computing Cluster, such data or service can show complicated form, like the data of various forms.The difference of data format cause between node mutual when cooperating in a different manner [4,6] handle data of different nature, reach the reasonable use of computational resource.The 3rd, the heterogeneity of node physical location.Because the formation of dynamic cluster is decided by data and service quality fully, and non-artificial prior definition systems stabilisation, this causes possibly comprising in the cluster from the big node of physical location difference.Suitable routing mode [7,8] is the high efficiency of whole cluster both, also can guarantee that node obtains fair service on the different physical locations simultaneously.The 4th, the heterogeneity of nodes ' behavior.The behavior of node in dynamic cluster also had any different.Some node can occur with computational resource contributor's form, and some then is to ask for and contribution is had both, and also some just asks for [9] merely.The dynamic cluster algorithm must be considered these characteristics, to guarantee the high efficiency of cluster.

Had the needed basis of dynamic cluster and to large-scale internet in dynamic cluster had understanding after, just can interdependent node organize with formation dynamic cluster.Based on the characteristic of above-mentioned dynamic cluster, it is different fully with the cluster under stable environment to set up the method for dynamic cluster.The drift net agreement is an important solutions wherein, and it is summarized as follows simply:

Foundation is based on the routing algorithm of community network [5].This routing plan is the resource finding algorithm that is based upon on community network or complex network [5] basis.Because community network is the topological structure that meets Internet resources distribution rule that computing node is set up through interconnected user on the network's social property in essence, utilizes this network just can accomplish the effective searching to computational resource rightly.When utilizing community network to carry out route search, relate generally to the combination problem of structuring and destructuring route technology.The outstanding feature of structuring route technology is in confirming finite time, to obtain route results to any resource (rare or popular), but such result is based upon under the strict constraints.And this condition current internet environment is difficult to be guaranteed.The destructuring route technology is based upon on the topological structure (being community network) that the Internet forms naturally, the internet environment that exists is not had any mandatory requirement.On such basis, will guarantee to obtain Search Results fast, necessarily requiring has deep understanding just can reach to concrete application and corresponding topological structure.But even so, unstructured searching is still not enough in the assurance of route fairness, and is also difficult to sharing of route pressure simultaneously.In view of the foregoing, can appropriately combine said two devices.Because have group in the community network, group is representing metastable computing environment, approaches the desired condition of structuring route.Like this, can take in whole the Internet, to carry out the destructuring route, and in group, carry out the structuring route.Through said method, routing algorithm has just been accomplished the popular resource of effective searching and has been guaranteed the route justice to scarce resource.

Set up the event-driven mechanism in the dynamic cluster.One of basis that dynamic cluster forms is exactly to be prerequisite each other alternately with the event-driven.Be that a node request connects, initiates transfer of data and closes connection or the like basic act and all accomplish through event-driven mechanism.This and traditional the Internet are that master's mode is completely different with the ballot.And after the basis of setting up dynamic cluster, this event-driven mechanism has just formed basic framework.Combine lightweight multicast mechanism efficiently again, just can form a message based event-driven mechanism.This mechanism can adapt to the characteristic of the Internet dynamic change, guarantees event-driven mechanism work under the reliable extensive environment of efficient stable.This machine-processed realization has improved the interactive efficiency between the Internet computing node, has avoided voting mechanism to the pressure of system's increase and to the system change torpid reaction.Event-driven has guaranteed that concern person can need not pay extra cost and vote before incident takes place; Simultaneously, after incident takes place, can react at once.The remote events driving mechanism becomes the basis that the reflection the Internet changes under the internet environment.The Internet is a system that constantly changes in essence, but this variation can't be reflected to the correlation computations node immediately under the support that does not have the remote events driving mechanism.In order to reduce computation burden, during current internet is mainly used in addition the refusal these variations are followed the tracks of.The present invention has solved this problem effectively.Dynamic cluster under the commercial computing environment in the Internet has own characteristic, need address the problem, and promptly adapts to the Internet dynamic environment, safeguards common commerce services, transmits lightweight data and management business etc.

(1) on common commercial object basis, sets up the dynamic cluster that adapts to internet environment.Dynamic cluster is that the computing node of a plurality of independent individuals or people's group control cooperates the distributed model that calculates formation in essence.Such computing node can add system at any time and also can log off at any time.The distributed system that they are combined is in the middle of constantly the jolting of producing therefrom.The dynamic cluster technology can be according to its characteristic of calculating between the concrete characteristic use node common commercial object, makes the relative equilibrium of this cluster in keeping dynamically, thereby effectively utilizes the computational resource on each computing node, reaches high performance purpose.Wherein relate generally to the relevant big data multicast technology of irrelevant big data multicast technology of maintenance to common purpose between the node, lightweight multicasting technology, sequential and sequential etc.These technology have mainly been used the Pareto optimality principle to carry out cooperating between node and the node, and have been combined the concrete characteristics in the large-scale distributed environment to form dynamic cluster on the basis that the individual who respects Control Node or crowd need.

(2) guarantee the correctness of common commerce services through the maintenance that global state is changed.In data-oriented internet, applications, the purpose that dynamic cluster forms is that common cooperation accomplishes transfer of data, and its thought is in this process, fully to share and utilize the node computational resource.Commercial calculating is different from the data-oriented internet, applications, except transfer of data, also has the calculating based on business logic.Because what the present invention taked is that the mode that whole business logics move alleviates the server cost, so the business logic that each computing node had is the same.Even so, concrete commercial calculating that each client is carried out is different, and related data also are different.When business logic acts on business data, can cause the variation of commercial state; And this variation must be forwarded on each node that data are relevant therewith at once, otherwise can cause the mistake of business process, causes economic loss.Can find out that for guaranteeing the correctness of Distribution calculation process, key is that global state changes the maintenance in the Distribution calculation node.For this reason, when forming cluster, should get up to form dynamic cluster to node organization with common business data; When one of them node caused data to change, all the other nodes can adopt multicasting technology to obtain such variation rapidly under the help of cluster, to avoid commercial mistake in computation.Though can not be equal to fully between business data and the commerce services, under most of practical situation, common business data just means common commerce services.So, can think to the maintenance of business data jointly in the dynamic cluster, be exactly maintenance to common commerce services.

(3) set up based on BT agreement lightweight data multicast mode.Current data-oriented internet, applications more and more develops to the heavyweight data.But for most of commercial calculating, correct and safety is most important requirement all the time.Business data generally can be expressed and transmits with the lightweight data.This and data-oriented internet, applications difference are very big, and most of commercial application can not cause big pressure to system in transmission at least.In the middle of the dynamic cluster under the commercial computing environment, transfer of data can effectively be reduced the cost that each node causes; When data were lightweight, transmission cost will be lower in cluster.Even so, because the behavioral characteristics of cluster is obvious, the lightweight reliability of data transmission has just become the problem of most critical; Especially all the more so concerning commerce is calculated.For this reason, the present invention has adopted based on the lightweight multicast system that improves the BT agreement.

(4) be based upon business way to manage on the lightweight reliable multicast agreement.Dynamic cluster and the migration of business logic data cause a calculating that is closely related on the whole to be distributed on a plurality of independent computing nodes.Under this condition, any demand that all can cause consistency maintenance to the modification of data.Here must under the support of reliable multicast agreement, adopt appropriate Managed Solution just can solve.Transaction management becomes the key that addresses this problem in the distributed system.But because the dynamic different with general distributed system, the transaction management among the present invention must be carried out modify.The back can be carried out illustrated in detail to this.

In the 4th step, set up based on the lightweight multicast system that improves the BT agreement.To the problems referred to above, the present invention has set up based on the lightweight multicast system that improves the BT agreement.Original BT agreement is a heavyweight data multicast agreement in essence; Its running environment is consistent with environment of the present invention.To the characteristic that business logic in the commercial systems for applications is accomplished based on the lightweight data basically, the present invention improves it, makes it be more suitable for the multicast of lightweight data.

(1) uses tracker monitoring dynamic cluster node state.Tracker in the BT agreement is the server that is used for collecting all node states in the potential dynamic cluster specially.In system of the present invention, this server can be realized through the method for arranging central server.For the lightweight multicast system, need the state of monitoring to comprise whether node is in the system, node whether to certain data pay close attention to, whether node be whether super node, node have obtained these data etc.These states all can play a key effect for the quality of incident multicast system.After node state is held in to system, could rationally share, improve multicast efficient and reliability lightweight multicast role pressure.Monitoring for node state can be accomplished through the form of participating in the commercial computing node proactive notification tracker that calculates in the present invention.In commercial application, the state that tracker need be monitored is used and will be lacked with respect to heavyweight, but shows as more concentrated.Mainly needing the state of monitoring is whether node needs certain business data and variation thereof, whether obtains related data and whether finish commerce services etc.Simultaneously, because business data itself has been main with lightweight, efficiency of transmission is high relatively, and this causes having at short notice a large amount of state variation of concentrating.

(2) tracker distributes the multicast task according to detecting the state equalization that obtains.In the BT agreement, the responsibility of tracker is that the node to new adding system returns the part of nodes of just in dynamic cluster, working, thereby makes new node carry out mutual with these nodes and add cluster, thereby progressively accomplishes the multicast of heavyweight data.In the present invention, the tracker major responsibility is by provide clustered node to be transformed into the balanced multicast task of distributing of the existing node of cluster for new node.Because the state that the lightweight data multicast relates to is limited, new node adds only needs the data that record needs on tracker to get final product.Lightweight data occupancy bandwidth is low; Each node can obtain related data fast; This node state that can cause in this process, preserving on the tracker changes frequent.For alleviating tracker for the burden of collected state variation monitoring too, can be employed in the node of confirming right quantity in all nodes, let these nodes at first on the lenient side the magnitude data produce node and obtain data.For commercial application, this quantity can require according to cluster total scale, instantaneity, specifically factor such as application characteristic and node characteristic is confirmed.Usually in such cluster, there is the minority super node, can be set to preferential node at first from data source acquisition lightweight data by these nodes; On average share the multicast task of residue node and return to tracker to correlation behavior information by these nodes again.

(3) confirm super node according to the formed historical information of line duration length.Tracker can be confirmed super node according to the formed historical information of node line duration length as lasting online server.Usually line duration is long more, and the possibility that becomes super node is big more.According to Complex Networks Theory [5], the online time span of minority node can be significantly higher than most of nodes.These nodes just become super node.Super node shows various features such as long, the stable and computing capability of line duration is strong, is the main computational resource of sharing tracker monitoring cost and accomplishing the multicast task.

(4) change node selfishness property and be contribution property.In the BT agreement, each node all is assumed to be with selfish mode and works, and is concerned about no longer after obtaining related data that promptly other node states will independently leave system; Simultaneously with other node reciprocal processes in also just satisfying just can be mutual with the other side under the oneself requirement prerequisite; Otherwise, can seek other nodes and carry out alternately.Such design is correct for dynamical system middle heavyweight data multicast.But for giving the little lightweight data of system pressure, just there is no need.In commercial application, require the business data can bamboo telegraph.At this moment each node should initiatively and by the node that returns on the tracker be got in touch after obtaining data, data is sent to the node that does not also obtain data.After obtaining data from super node, with respect to most ordinary nodes, they are for average relatively on the disposal ability of lightweight data, and multicast pressure just balancedly is distributed on each node like this, has guaranteed the instantaneity of transfer of data simultaneously.Because multicast is carried out in no obvious topological structure, the behavioral characteristics of node is dropped to minimum for the influence of multicast quality.

(5) guarantee the order of lightweight data in multicast process.For business data, must guarantee that the order of business data in transmission course is to guarantee the correct completion of business logic.On the Internet of strong dynamic, more need carry out strict maintenance to the order between the data.Realize above-mentioned purpose, tracker has played coordinative role.When a node is attempted to send data, at first to send request to tracker; Tracker sorts according to the request sequencing to all nodes that need send data in the cluster; Multicast is carried out according to this sequencing successively, and promptly after the node data of confirming to come the front sent successfully in whole cluster, the node data that comes thereafter sent again.Such way can cause whole system because excessively synchronously and decreased performance.Also can take to carry out asynchronous transmission then to being carried out serial number by sorting data.When each node was handled the data that obtain simultaneously, if find that numbering is discontinuous, then can judge had data not receive, can get in touch with data source and obtain related data.This method can be raised the efficiency.

In the 5th step, set up based on distributed independent business management strategy.The present invention is based upon under such prerequisite, and it is few that the related business logic of commercial server calculates resource requirement; But when facing a large number of users and visit simultaneously, it is in short supply and have to adopt technology of the present invention that computational resource can appear in commercial server.The client of each visit commercial server can be after the business logic migration, and this business logic of execution makes an explanation in this locality.In addition, be independent of each other in business activity between each client, do not exist to be distributed in different clients but the common situation that forms a business; This characteristic makes that traditional distributed transaction management method obtains simplifying, and promptly any one business all only is present on the computing node.Under these conditions, the present invention proposes distributed independent business management strategy.This transaction management method also will guarantee the balance of calculating pressure except the demand that will satisfy transaction management itself.

(1) original commercial server is participated in transaction management as the transaction management coordinator.Though each business independently is present in respectively on the computing node, they all are based on common business data, and this causes the possibility of commercial results mistake in the concurrent implementation of business logic.Because each node all carries out business logic in concurrent independence based on corporate data, this just require these independently computing node have common state.This common state is exactly a sequential.For the maintenance of affairs, basic principle is exactly to make the concurrent operations that possibly clash change into the serial operation that meets sequential.In distributed system, the sequential of each computing node is skimble-scamble.For this reason, the original commercial server role that just can serve as this coordinator is unified to guarantee sequential.In addition, original commercial server also relates to some other work, comprises write operation business node statistics etc. like entering.Because computing node has been shared the pressure of a large amount of commercial calculating and transaction management, the calculating pressure of as the coordinator, accomplishing above-mentioned work can be little, and general computing equipment can bear.

(2) participate in the transaction management cost for alleviating commercial server as the coordinator, dwindle the timing management scope through only monitoring the transaction write operation.The essence of transaction management is exactly to guarantee data consistency in the business logic correct execution.If but letting original server bear too much transaction management pressure, and will cause network delay to increase, systematic function descends.For this reason, must manage to dwindle the transaction management scope.Through a large number of users is participated in being the understanding of characteristic business logic, can obtain phenomenon, the business of a large number of users with in do not comprise write operation.In addition, as extensive commercial system, even write operation is arranged, the corporate data chance that they are faced at synchronization is also little.Like this, original server only needs in relatively little scope, to carry out transaction management: only need each node to notify original commercial server the business executing state that self comprises write operation.Simultaneously, original commercial server is added up these states, is registered by the affairs of write data with having jointly chronologically, so just kept the computing node that affairs are conflicted might take place.When a computing node will carry out write operation, can notify original commerce server; Original server is responded interdependent node and whether can be write or postpone the products for further notice at once according to transaction management SECO principle.

(3) accomplish the recovery after affairs are abandoned through the control of transaction write time sequential routine and to the monitoring that node logs off.Principle according to transaction management can be clear and definite, as long as the write operation of affairs carries out after promising to undertake early than the affairs of its generation or abandoning, and the state before when it is abandoned, just can abandoning to it according to the historical record restore data.Because the state that comprises each node write operation affairs has been held in original commercial server strictness, this has just guaranteed when affairs are abandoned, and still can recover and keep data consistent and commercial calculate correct.Carry out commerce services owing to the present invention is based on the client of individual's control, so dynamic will be higher than the conventional commercial system far away.For this reason, judge whether a computing node leaves system and just become the recoverable key technology of assurance affairs.If a node normally leaves, original server can normally be carried out recovery operation so; If but improper the withdrawing from of node can cause original server can't judge affairs are accomplished or abandoned to this node whether usually.For addressing this problem, require among the present invention in its neighbours, periodically to send heartbeat data at the node of the enterprising line transaction registration of original server.If its neighbours do not receive heartbeat data in the fixed cycle, this can think that this node is improper and withdraws from, and promptly abandons affairs; Afterwards, these neighbours understand the original commercial server of proactive notification to accomplish the data necessary recovery.

In the 6th step, set up personalized commerce services based on the commercial dynamic cluster that a large number of users participates in forming.Personalization in the business system mainly is from mass data or service, to confirm user's request effectively, directly satisfies the special system of its demands of individuals for it provides.In current business computing system, this purpose realizes difficulty or poor effect.Its subject matter is that user personality is difficult for holding; Related algorithm is inaccurate and efficient is low.This has had introduction in front.The personalized realization of the present invention relies on dynamic cluster and accomplishes.Each ripe dynamic cluster all is a sub-set of original commercial server business logic, has possessed a part of data of whole commercial services.When new user capture original server, original server can be forwarded to the associated dynamic cluster to this request, makes it obtain related service.Since in the data on each dynamic cluster in associated user's wish control down through ask buffer memory acquisition progressively to server, can think that a dynamic cluster is exactly the computing system that satisfies a kind of individual demand.Original server can be asked the commerce services of judges care according to the user; Forward the data in view of the above again on the relevant cluster and just can accomplish personalized service.Need to prove that dynamic cluster says that the personalization that possesses is without any need for extra intervention; Self has just possessed individualized feature each dynamic cluster; This is that nature forms, and is the embodiments of intellectual resources in computing system.Have only a large number of users to participate in large-scale distributed system and just possess these characteristics.

In the 7th step, the dynamic cluster that utilizes commerce services to form is realized the heavyweight data, services.It is higher that heavyweight data in commercial the calculating can make commerce services itself improve quality more.When the heavyweight data occurring, the present invention can utilize dynamic cluster to accomplish the transmission of heavyweight data.Dynamic cluster under this state is except being organized in together based on common commerce services, and more resources spends on the common heavyweight data that need of transmission.As a rule, these two kinds of clusters that form based on various computing are highly to overlap; This is that user's request convergent physical feature in dynamic cluster causes.Therefore, can not bring influence to other clusters.

In the 8th step, realize the commercial computing environment of unlimited scale through utilizing the client computational resource.The formation of the commercial computing environment of unlimited scale is key issue for the large scale business service system.The formation of unlimited scale computing system depends primarily on the retractility of system.Its potential scale of system that possesses strong retractility also can be big; Vice versa.Because the present invention has taked the strategy of business logic and relative commercial data migration client thereof, make to calculate on the computing node of all request commerce services and carry out; It is little that simultaneously most of commerce are calculated potential amount of calculation, in fact becomes the trend that an entire system computational resource goes up not down along with the increase of user capture amount; Effective in addition lightweight Data Transport Protocol makes the network operation burden reduce.This has caused business system still can deal with along with scale increases even can possess potential customers' end that stronger ability is admitted bigger even unlimited scale.And the conventional commercial system can move closer to the resource limit along with the scale increase; This and business system of the present invention form distinct contrast.From practical application, the present invention realizes the target of unlimited scale except helping business system, has also saved the input of conventional commercial system greatly, also for the commerce services of medium-sized and small enterprises low-cost computing platform is provided.

Good effect of the present invention is:

(1) reduction is based on the input of the Internet commercial server computational resource.Primary and foremost purpose of the present invention is exactly to reduce the huge input cost of large scale business.Through the use of the present invention's technology, the commercial server in the large scale business system only needs can reach than higher service quality under the current high investment situation far below current input.Can save the design cost of computing equipment, bandwidth and reduction server software through user of the present invention.

(2) realize the large-scale commercial applications computing system under the limited computational resource condition.Though use of the present invention can reduce the input of server computational resource greatly, do not hinder a business system under such condition, to deal with potential a large amount of clients' visit.When system scale enlarged gradually, system still can stably receive user's request, and all requests are responded rapidly.Under the prerequisite that server resource suitably guarantees, native system can reach the effect of unrestricted access.This application system institute that is based on current commercial server is insurmountable.

(3) personalized service of assurance business system.Large scale business for based on the Internet is used, and this is a tangible demand.The present invention can so that the user in visit during business system, can be only mutual each other at the Service Source relevant with its sexual demand, seek in the large scale system and the cost of self-demand match service and be reduced in.

(4) in commerce services, guarantee the heavyweight data transmission performance.Increasing business system begins to provide the heavyweight data, services.This has increased service quality and user experience effect greatly.Commercial Internet Server among the present invention can provide the more heavyweight data, services of high-quality, and can requirements at the higher level not arranged to the input of computational resource.

(5) foundation is based on the commerce server of cheap computing equipment.For some medium-sized and small enterprises, there are not enough financial resources to put in the middle of the Internet commercial application.The commercial server that the application of the invention relates under the situation of insufficient investment, can just can construct a high-performance commercial server through using ordinary individual's computing equipment (like personal computer) and general commercial bandwidth (like ADSL).If additional related is technological, its effect even can reach or near the large scale business application results.

(6) reduce the additional management cost as far as possible.Though the present invention brings above-mentioned variation for the Internet commercial application,, also can bring some problems because the present invention has transformed some basic structures of current internet.This problem shows the most outstandingly in the middle of commerce is calculated.Its subject matter is embodied in the management cost and improves.The present invention has also proposed relevant solution, guarantees when bringing positive the variation, can reduce the additional management cost as far as possible.

Description of drawings

Unlimited scale commercial server workflow.

Embodiment

In the middle of practical implementation of the present invention, can run into following several kinds of concrete situations usually.Through using technology of the present invention, can guarantee that these application accomplish unlimited scale and high-performance.

The first, set up small-sized ultra-large business system through personal computing devices.The enforcement of current internet business system gives no thought on personal computing devices sets up.Personal device is defined as computational resource limited, no fixed internet address own and the low computing equipment of bandwidth here.On current internet, such equipment occupies great majority in the whole node that is connected on the Internet.Why the current internet business system is not considered to utilize personal computing devices to set up business system, mainly is because such equipment can not provide enough computational resources.Based on the means of setting up of conventional commercial system, personal computing devices can't satisfy the client to the basic demand of commerce services after scale enlarges gradually.By the present invention, can on personal computing devices, set up small-sized ultra-large business system.Here it is limited that " small-sized " refers to the service that business system itself comprised business data simple, that relate to, is not the potential small scale of this business system.Here " small-sized " is by personal computing devices storage ability and individual commercial supplier's characteristics determined.Because memory space is relatively limited, its business data that comprises is restricted as personal computing devices; Simultaneously, the commerce services that provides of personal user is concrete and single usually.But such system might have potential extensive.Realize above-mentioned target, the practical implementation step is:

(1) at personal computing devices deploy business logic and business data;

The client of (2) having a mind to enjoy this commerce services is installed unified business logic interpreter;

(3) accept on the internet to migrate to this client from the commerce request of client and business logic;

(4) carry out buffer memory according to user's request to the client transmissions business data and in client;

(5) when the client scale increases, the commerce services classification of enjoying according to the user forms dynamic cluster; The commerce services classification is meant concrete commerce services, as sells book, sells electrical equipment etc.;

(6) in dynamic cluster, carry out state-maintenance and business management through multicast;

(7) under the situation that personal computing devices allows, the heavyweight data, services that run-of-the-mill is provided for the user is with enhanced quality-of-service.

The second, set up towards the ultra-large business system of personal user through large-scale service equipment.Except utilizing personal computing devices to provide the commerce services,, can drop into commerce services kind that much higher computational resource guarantees to enrich and a large amount of business datas for the enterprise that is devoted to provide the large scale business service.But to set up the method that the large scale business service system taked different with tradition, utilize the computational resource total amount of input required for the present invention low, simultaneously also difference of major function.In traditional the Internet business environment, all commercial calculating and corresponding management task all concentrate on these equipment; For business environment of the present invention, the computational resource major function of input is to manage but not directly commercial calculating.The computational resource that why need Duo relatively than personal computing devices is in order to store more business data, deal with more substantial management work and high-quality heavyweight data, services etc. to be provided.

(1) at large-scale service equipment deploy business logic and business data;

(5) when the client scale increases, the commerce services classification of enjoying according to the user forms dynamic cluster;

(7) provide high-quality heavyweight data, services with enhanced quality-of-service for the user.

The 3rd, for the enterprises and individuals business environment is provided through large-scale service equipment.In the Internet commercial application; Except providing business logic and business data to provide the commerce services for the terminal use; Also having a kind of situation is for the enterprises and individuals provides business environment, thereby helps these enterprises and individuals to come for the user commerce services separately to be provided by such business environment.Utilize traditional Internet technology to realize that above-mentioned application requirements ISP bears all computational resources.These resources comprise the primary commercial logic, can adapt to business logic, computing capability, memory headroom, lasting memory space and bandwidth or the like.Except concrete business data, nearly all resource is all dropped into by the supplier.Also have a situation to be, the supplier of business environment can also prepare succinct description interface and supply the user to express its unique business logic; Simultaneously, have the business logic that the logical interpretation device is responsible for carrying out the user in the business environment.But which kind of situation no matter all requires enough computational resources to deal with all users' calculating pressure to the supplier of business environment.Be consistent with the commerce services of directly facing the terminal use in this.For the calculating pressure that reduces the business environment supplier and enlarge business environment as far as possible and say the userbase that holds, can adopt technology of the present invention.Utilize the present invention, the supplier of business environment only need bear the primary commercial logic, can adapt to business logic, business logic descriptive tool, business logic interpreter and low computational effort resource get final product.Reaching the reason of this target, is being in charge of but not directly commercial calculating because of the business environment service under the present invention supports equally.

(1) on large-scale service equipment, provide main enterprises or individual to carry out the related business logic of commerce services;

(2) except concrete business logic, the description and the interpreter thereof that can adapt to business logic, business logic can also be provided;

(3) accept enterprise or individual request, and require to be its configuration relative commercial logic according to it to business environment;

(4) require the user that business data is provided;

(5) accept the request of secondary user's and move corresponding business logic to it;

(6) according to the request of secondary user's at subscription client buffer memory business data;

(7) between the client of secondary user's, form dynamic cluster;

(8) carry out state-maintenance and business management in dynamic cluster inside through multicast;

(9) carry out the heavyweight data, services in dynamic cluster inside.

The 4th, set up the enterprises business environment through small-sized service equipment.Usually also there is a business environment an enterprises.Enterprise carries out enterprise internal management through this business environment.The difference of this environment and true internet environment is mainly reflected in that environment is safer, Internet resources are abundanter and more stable.In this case, the present invention is not so good as under true internet environment obvious to the transformation effect of system.If but be utilized in the present invention in the enterprises business environment, still can effectively reduce requirement, and obtain the support of stronger heavyweight data, services enterprise's business environment resource.In such environment, the basic mode that the present invention uses does not change: main guiding theory remains has calculating center service equipment to transfer to client, and center service equipment only is responsible for coordinating and management role.Different is; Because Intranet aboundresources; The contact factors such as administrative mechanism that cost is low and enterprises is intrinsic between the computing node; The system that causes the enterprises node to form tends towards stability, and at this moment the mechanism based on the lightweight multicast can adopt simple tree topology structure to get final product.In addition, when system scale was little, the heavyweight data, services also can be taked tree-shaped or the forest topological structure is that main form is carried out.Here mean in fact that setting up of dynamic cluster is convenient, and need not such complicated mechanism under the large-scale internet environment.In any case, be transferred to client owing to calculate, make just can obtain than the current centralized stronger calculated performance of mode at the small-sized service equipment of the use of enterprises.

(1) at enterprises small-sized service equipment deploy business logic and business data;

(2) on the client of enterprises participation commerce services, unified business logic interpreter is installed;

(3) on Intranet, accept to migrate to this client from the commerce request of client and business logic;

(5) when the client scale increases, the commerce services classification of enjoying according to the user forms the cluster based on the tree topology structure;

(6) in above-mentioned cluster, carry out state-maintenance and business management through multicast;

(7) in above-mentioned cluster, provide high-quality heavyweight data, services with enhanced quality-of-service.

The 5th, set up business environment between enterprise through small-sized service equipment.In current commercial computing environment, also there is collaborative calculation mode between a kind of enterprise.Different enterprises are because operational needs have also produced dependence each other, thereby formation is based on the commercial application of enterprise calculation resource.More general situation is that each enterprise also has a large amount of clients; It is the center that these clients provide commerce services with an enterprise; And have between the enterprise because commercial relations form association.Under such complicated situation, can adopt the front second way to calculate between an enterprise and its client; And between enterprise, owing to business is stable each other, computational resource is abundant, this can take the inner mode of similar enterprises is that the 4th kind of situation solves the calculating pressure problem.

(1) accomplishes the commerce services between enterprise and its client with the second way;

(2) accomplish the commerce services between the enterprise with the 4th kind of mode.

List of references:

【1】Vinton?Cerf；Specification?of?Internet?Transmission?Control?Program；RFC?675，December?1974

【2】Rosenberg?J.，et?al；STUN-Simple?Traversal?of?User?Datagram?Protocol(UDP)Through?Network?Address?Translators(NATs)；RFC?3489，March?2003

【3】Ananth?Grama，et?al；Introduction?to?Parallel?Computing；Second?Edition，Benjamin/Cummings?Publishing?Company?Inc，1994，ISBN：0-201-64865-2

【4】Cohen?B.；Incentives?Build?Robustness?in?BitTorrent；in?Workshop?on?Economics?ofPeer-to-Peer?Systems，Berkeley?USA，May?2003

【5】Newman?M?E?J.；The?Structure?and?Function?of?Complex?Networks；SIAM?Review，2003，45，Page(s)：167-256

【6】Vlavianos?Aggelos，et?al；BiToS：Enhancing?BitTorrent?for?Supporting?StreamingApplications；INFOCOM 2006，25 ^th?IEEE?International?Conference?on?ComputerCommunications?Proceedings，April?2006，Page(s)：1-6

【7】Rowstron?A.，Druschel?P.；Pastry：Scalable，Decentralized?Object?Location?andRouting?for?Large-Scale?Peer-to-Peer?Systems；in?Proceedings?of?the?2001?ACM?SIGCOMMConference?on?Applications，Technologies，Architectures，and?Protocols?for?ComputerCommunication(SIGCOMM’01)，San?Diego，CA，August?2001，page(s)：247-254

【8】Gribble?S.D.；Measuring?and?Analyzing?the?Characteristic?of?Napster?and?GnutellaHosts；Multimedia?Systems?Journal，9(2)，August?2003，page(s)：170-l84

【9】Adar?E.，Huberman?B.A.；Free?Riding?on?Gnutella；First?Monday，Internet?Journal，Oct.2000，Available?at?http://www.firstmonday.dk/issues/issue5?10/adar/index.html

Claims

1. the commercial service system works method of a Internet large-scale application environment the steps include:

1) adopt TCP/IP peer protocol pattern to make computing node have the ability of the request of reception and transmission request; Said node comprises the client and server end;

5) in above-mentioned dynamic cluster, carry out state-maintenance and business management through multicast, wherein:

The multicasting method of said dynamic cluster is:

A) use tracker monitoring dynamic cluster node state; Said state comprises whether node is in the system, node whether to certain data pay close attention to, whether node be whether super node, node have obtained these data;

B) tracker distributes the multicast task according to the state equalization that monitoring obtains, and its method is: at first tracker confirms that according to the formed historical information of node line duration length the node of right quantity is a super node; Let then these super nodes at first on the lenient side the magnitude data produce node and obtain data; Last said super node is on average shared the multicast task of residue node and is returned to tracker to correlation behavior information; Wherein, tracker is coordinated said super node and is carried out the multicast task, and its method is: a) when a super node is attempted to send data, at first will send request to tracker; B) tracker sorts according to the request sequencing to all nodes that need send data in the cluster; C) tracker carries out asynchronous transmission then to being carried out serial number by sorting data; When d) each node was handled the data that obtain, whether judgment data lost, if find that numbering is discontinuous, then got in touch with data source and obtained related data;

The business management method of said dynamic cluster is:

A) the server end concurrent operations that possibly clash is changed into the serial operation that meets sequential, and its method is: at first node is with the business executing state announcement server end that self comprises write operation; Server end is added up the transaction write mode of operation then, is registered by the affairs of write data with having jointly chronologically; Last server end is responded interdependent node and whether can be carried out write operation at once or postpone the products for further notice according to transaction management SECO principle;

B) accomplish the recovery after affairs are abandoned through the control of transaction write time sequential routine and to the monitoring that node logs off; Its method is: the node that carries out the affairs registration at server end periodically sends heartbeat data to its neighbours; If its neighbours do not receive heartbeat data in the fixed cycle; This can think that this node is improper and withdraws from, and promptly abandons affairs; Afterwards, these neighbours understand the proactive notification server end to accomplish the data necessary recovery; If node normally withdraws from, server end is normally carried out recovery operation.

2. the method for claim 1 is characterized in that adopting the SSL method that said business logic is moved to client.

3. the method for claim 1 is characterized in that the transmission method of said transmission business data is: adopt the method for buffer memory progressively business data to be moved to client from server end.

4. the method for claim 1 is characterized in that setting up said dynamic cluster according to the drift net agreement; Adopt the Pareto optimality principle to carry out the cooperation between the node in the said dynamic cluster.

5. the method for claim 1 is characterized in that when new user access server, and server forwards the data to the associated dynamic cluster more in view of the above according to the commerce services that the user asks judges to be concerned about, makes it obtain related service.

6. a commercial service system that adopts the said method of claim 1 comprises server end and client, it is characterized in that said server end deployment business logic interpreter, business logic and business data; Said client deployment has the business logic interpreter; Said server end is connected through network with said client.