CN100397349C - Real composite objects for providing high availability of resources on networked systems - Google Patents
Real composite objects for providing high availability of resources on networked systems Download PDFInfo
- Publication number
- CN100397349C CN100397349C CNB028234367A CN02823436A CN100397349C CN 100397349 C CN100397349 C CN 100397349C CN B028234367 A CNB028234367 A CN B028234367A CN 02823436 A CN02823436 A CN 02823436A CN 100397349 C CN100397349 C CN 100397349C
- Authority
- CN
- China
- Prior art keywords
- composite factor
- node
- utility
- framework
- service
- 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.)
- Expired - Lifetime
Links
Images
Abstract
A composite resource is established that includes a plurality of members. Each of the members is capable of providing a comparable service. A coordinator monitors a state of each member of the composite resource. A component requests the service from the coordinator. The coordinator arranges for the service to be provided to the component by a particular member of the composite resource. When the particular member ceases to be active, the service is automatically provided to the component by another member in the composite resource. A state of the composite resource is maintained independently of the state of each member in the composite resource.
Description
Application reference
The application requires the U.S. Provisional Patent Application 60/334 of submission on November 30 calendar year 2001,349 right of priority, its title be " fast quick-recovery incident; virtual composite object; and the enables/disables attribute (RapidRecovery Events; Virtual Composite Objects; And Enable/DisableAttributes For Simple and Composite Resources For Use In A HighAvailability Framework) that is used for the simple and composite factor of high availability frame ", the inventor is Colrain etc.; In addition, the application also requires the U.S. Provisional Patent Application 60/426 of submission on November 15th, 2002,587 right of priority, its title are " the virtual composite object (Virtual Composite Objects for Providing High Availability ofResources on Networked Systems) that high available resources are provided on network system " (attorney docket number: 50277-1968).Above-mentioned all being hereby expressly incorporated by reference as basis for priority in first to file.
Technical field
The present invention relates to the resource management of network system.Especially, the present invention relates to be used for the true composite object of network system for managing.
Background technology
In the node cluster, the availability of resource is an important consideration.Therefore, in order to increase the reliabilty and availability of this cluster, need make some resource redundancy in the cluster effective.When a node or resource failed, exist certain mechanism to make another node use similar resource usually.
The resource of node is piled up by present system.In storehouse, resource develops its dependence each other.Therefore, the middle layer resource can depend on the resource of lower level, and application or other higher-layer programs then depend on the resource of lower level.In the past, when taking place when losing efficacy, the dependence of management resource has problem in storehouse.
A kind of scheme is to rebuild the storehouse of the node that lost efficacy on another node fully.Even have only a resource failed, this scheme still rebulids whole storehouse in other place usually.Therefore, will cause system providing taking place delay on the redundancy services ability of the node that lost efficacy in the resource failed of a node.This delay (time) is to such an extent as to very long usually quality and the availability that the service that provides from the node cluster is provided.
The another kind of scheme of attempting in the past is to use framework to realize switching between the node when lost efficacy taking place.In this class scheme, the framework of node is rebuild the storehouse of node in the interconnected cluster in the place of the generation of losing efficacy.In this system, the availability of node cluster depends on the response of framework, and framework is in the ability of the storehouse of different node rebuild resources.
Description of drawings
The present invention is described by way of example rather than in the mode of restriction with reference to accompanying drawing, in the accompanying drawings, and same Ref. No. indication components identical, wherein:
Fig. 1 shows and uses composite factor that service method is provided on the node cluster;
Fig. 2 shows the ultimate system of setting up and use true composite factor on the node cluster;
Fig. 3 shows another ultimate system of setting up and use virtual composite factor on the node cluster;
Fig. 4 shows the method based on a plurality of members of service level management composite factor;
Fig. 5 shows the system that is configured to detect immediately interested incident;
Fig. 6 shows the basic fundamental of management composite factor when a member was lost efficacy;
Fig. 7 shows the another kind of technology of using composite factor after node is disabled; And
The block scheme of Fig. 8 shows the computer system that example of the present invention can be realized thereon.
Embodiment
The invention describes the method and apparatus that the high availability resource is provided on network system.In the following description, for illustrative purposes, a large amount of concrete details have been proposed, so that thorough understanding of the present invention is provided.But obviously, the present invention can realize under the situation of these details not having.In other example, well-known structure and equipment are presented in the block scheme, to avoid that the present invention is caused unnecessary bluring.
General general introduction
According to an example, the resource on the framework (framework) has been set up the composite factor (composite resource) that comprises a plurality of members.Each member can both provide comparable service.The state of each member in the framework agent monitors composite factor.Assembly for example uses resource such as network ip address and detectaphone to ask service.Resource, for example network intercepting device etc. arranges the one or more members in the composite factor to provide service to this assembly.When this special member is stopped action, just provide this service to this assembly automatically by other member in the composite factor.The state of composite factor is independent of the state of each member in this composite factor to be safeguarded.
According to another example, by on the node cluster, setting up the composite factor Management Unit on framework that comprises a plurality of members.Each member can be used to provide and can relatively serve.One distribution type logic (circuit) resides on the cluster and on the node.This logic can be carried out by the framework agency.This logic (circuit) still works in the time of can being configured to any one node failure in a plurality of nodes continuously.Use this logic (circuit) to collect each member's status information.Determine the service level of this service according to each member's state.This distribution type logical circuit arrange one or more members provide service to just in requesting component, up to being on close level till current unavailable service of this service.
In one embodiment, distributed logic (circuit) makes each member's executable operations of composite factor, to start this member, stop this member or to check the composite factor state.Other operation that distributed logic may be carried out comprises the operation that relates to the member condition that is enabled or forbids.
In another embodiment, initiatively interested incident is offered the framework of high availability.In the framework that has a plurality of assemblies to carry out, this assembly periodically poll detects whether interested incident generation is arranged.Also set up watch-dog for one or more assemblies.After assembly made interested incident generation, this watch-dog was initiatively given this framework interested event notification, and need not wait for this first assembly of framework poll.
In another embodiment, set up composite factor Management Unit framework by the member on residing in different nodes.Each member can both provide comparable service.Framework may receive that message forbids the member in this group node any one.As the response that receives message, this member is not re-used as the candidate of composite factor.This disabled member is equivalent to the member of off-line.
Term
" composite factor " is the set by a plurality of member compositions that utility is provided.This service can be corresponding in the multiple function of certain member's execution." homogeneity (homogeneous) " composite factor correspondence has on the function or much at one member's composite factor on the structure.In addition, composite factor can also have " heterogeneous (the heterogeneous) " attribute that has difference on member's function or the structure.Therefore, the service that provides of each member of composite factor needs not to be identical or of equal value.
" member " is the resource on the network system node, can be used to carry out certain function or the service of composite factor definition.Member's example comprises that application, application example and other enable and make things convenient for the resource of intra-node and PERCOM peripheral communication.
" framework " used herein is meant that the assembly that service is provided in the system is used for making it to serve software or other logic of the delay minimization that the recipient experiences when breaking down.Framework can be distributed, so that a plurality of nodes on diverse location can be shared.
Statement " essentially equivalent " means that a comparison point (quality, function or result) and another comparison point are identical or slightly different with another comparison point.If the function that each resource is performed and another resource have 90% correlativity, then claim this two kinds of resource essentially equivalents.
In whole the application, term " first ", " second ", " the 3rd " are as the identifier of distinguishing disparity items.When these terms were used as the identifier of project, this term did not have the seasonal effect in time series implication.Such as, term " first assembly " is meant one and is different from the assembly of " second assembly ", and needn't be meant more Zao than second assembly in time assembly.For another example, the member that first member can be corresponding movable, second member then can be corresponding to the members as the standby member of one or more other active members.
Use composite factor that service method is provided
Figure 1 illustrates and use composite factor in the node cluster, to provide service method.This node cluster can be a part that comprises the network system of miscellaneous equipment and assembly.In one embodiment, composite factor forms by the member who resides in the node cluster.Fig. 2 and Fig. 3 provide the example of the system that comprises the composite factor in the node cluster.
In step 110, the member from reside in the node cluster sets up composite factor.Can be interrelated between these members, so that each member provides the assembly of comparable service to this service of request.This member can be tradable, so that this comparable service can be provided by any one member among a plurality of members.Member's 26S Proteasome Structure and Function is determined by the service that composite factor provides.Such as, the database application service can have the composite factor that forms by for the member of database application example.
In one embodiment, the service that provides of composite factor has the reflection (image) of triangular web.Therefore, client, user or other are asked the complexity that the entity of this service has been hidden provides this service required.This service allows the entity of request to benefit from the redundancy section that is arranged on the node cluster.Provide the member of this service can switch to another member in the composite factor pellucidly, even this member resides on the different nodes.
When the member was just providing the function of this service, composite factor also made the member more effectively be managed.Such as, this composite factor can be used to plan the maintenance of clustered node, and need not plan the member in the composite factor separately.Therefore, when a member or member node are removed when safeguarding, the service that this node provides can be transparent and automatically provide by residing in other member in another node.
Can set up composite factor various dissimilar services are provided.The example of COS comprises that database service, functional service, data rely on service and pre-connection service.The function that database service is realized in the time of can be corresponding to the example of carrying out database server.Functional service can refer to grouping concrete function together when carrying out database application.Data rely on service and refer to send the process of session to service and resource.A specific session can state that (declare) uses a specific resource, and these data rely on service and then send this session to the position that can receive this service.Corresponding other resource using or when losing efficacy, support other service of pre-connection service.
According to an embodiment, the user provides required instruction and/or data to set up composite factor.The basis that forms composite factor is the service of user expectation.Such as, the user can import the example of the optimum position of row service corresponding to be received.The user can also refer to fix on the additional resource of other position, and its service is provided as supporting.In this example, the example of the assembly of composite factor by being used to provide this service forms, thereby each example all is the member of this composite factor.
In step 120, a member from composite factor provides service to requesting component.Requesting component can be the client terminal outside the resident node of member.In addition, requesting component can reside in the node, and needs this to serve to carry out.Therefore, an embodiment provides based on the choice of location member who compares resource to be supplied who asks assembly.This position can be transparent to middle level application or client terminal.
In step 130, provide the member of this service to be detected as to become service can not be provided.The corresponding various situation of this possibility, when node failure, this member will be lost efficacy.In addition, member itself may be lost efficacy, and node also remains on operable state.As another example, node and member are stopped, with the maintenance of planning.
In step 140, composite factor reappraise self determine its state with and member's state.This determines to comprise whether other member who determines in the composite factor can be used for providing service.Therefore, in one embodiment, this at least one other member who determines that requirement will form this composite factor is identified as movable or the exercisable state of being in.In addition, need determine that also this member is whether available or have the ability to provide this service.
In addition, lost efficacy, whether satisfied the specific radix of this composite factor in case step 140 can be determined this member that service is provided.Although a general example supposes that the member's radix in the node cluster is " 1 ", other embodiment can specify higher radix.Such as, this radix can be " 3 ", this moment, three members in the composite factor must be movable for thinking movable composite factor.
If another member who forms composite factor that determines who does in step 140 can provide this service, provide the service of this requesting component reception in step 150 so from this member.In one embodiment, from for the user being the service that transparent reposition continuation transmission offers a user.Be sent to before one or more other provide the member of this service in service, the user can receive the interruption of service.In addition, when the member that service is provided was lost efficacy, service transmitted between other movable member, so that the user receives continuous service.In such embodiments, the member of composite factor can be homogeneity in itself.Therefore, offer the request assembly service can not interrupt.In addition, providing other assembly on the node of service can also depend on this service.When the performance of these assemblies when a member is sent to other member can not be subjected to tangible influence.Therefore, embodiment can be provided this service continuously as described in Figure 1, and does not postpone when transmitting between the member of composite factor.
In the composite factor that is formed by heterogeneous member, the service that each member provides can be different qualitatively or in essence.But, can provide continuously, and the general utility functions of the service of this composite factor of definition may be provided to a certain extent pellucidly.
If in the member who determines the formation composite factor of step 140, do not have other member that this service can be provided, so in step 160, this service stopping that this composite factor provided.In one embodiment, composite factor only is only available when the minority member of certain member or appointment can restart.
Use the system of composite factor
Fig. 2 shows the ultimate system of setting up and use true composite factor on the node cluster.This composite factor provides service for other assembly, and these assemblies can reside in the inside or the outside of node cluster.In the embodiment that describes as Fig. 2, this composite factor is called as " really ", and reason is that this composite factor has physical interdependence at least one node in the cluster.Other example, described such as Fig. 3, so-called " virtual " composite factor is provided, reason is that this composite factor does not all have physical interdependence to any independent node.
In one embodiment, system 200 comprises first node 210, Section Point 220 and the 3rd node 230.This first node 210 comprises first resource 212, first member 214 and first assembly 216.Section Point 220 comprises second resource 222, second member 224 and second assembly 226.Similarly, the 3rd node 230 comprises information resources 232, the 3rd member 234 and the 3rd assembly 236.Can set up this first member 214, second member 224 and the 3rd member 234 to form real composite factor 240.
Framework 260 is present in first node 210, Section Point 220 and the 3rd node 230 each.Although the form that framework 260 is described as to reside in the part in the different nodes exists, an embodiment provides framework 260 to form the unity logic entity that is present on all these nodes.Framework 260 exists on each node with framework agency 266 form, and on behalf of the resource of local node, it carry out function.Framework 260 can be included in and pile up the application of failure node and the function of resource on other node again.In one embodiment, framework 260 run time versions are set up composite factor 240, and provide service to other assembly.Framework 260 and/or telegon 250 can use register 262 to store the member's of relevant composite factor 240 status information.This status information can indicate the member of this composite factor 240 to be in activity or inactive state.Register 262 can be the part of framework 260, and perhaps in another case, register can reside in the outside of framework.
True composite factor
In system as described in Figure 2, composite factor 240 is considered to " really ", because this composite factor is regarded as direct control by framework 260 and depended on a node in this system.This specific node of being regarded as the position of composite factor 240 by framework 260 is arbitrarily, may take place to change several times in the process of these composite factor 240 operations.Therefore, this node 210,220 and 230 is equal to, and reason is that each node can be counted as comprising this composite factor 240 in certain particular moment.
In Fig. 2, illustrate that by telegon 250 composite factors 240 are counted as being present in the mode on certain node.In one embodiment, telegon 250 is actually the code of carrying out with assessment composite factor 240 (code) on framework 260.In fact telegon 250 can be framework agency and the representative that is distributed in the resource on all nodes.But when framework 260 was carried out the instruction of the corresponding function that will be carried out by composite factor, this function (and composite factor 240) was counted as carrying out on a node.
Therefore, " truly " composite factor 240 be node independently, the specific node that this composite factor relied on be arbitrary decision and may change.If in certain particular moment, be counted as the resident node failure of composite factor 240, then this composite factor restarts on another node pellucidly.Thereby the member of composite factor 240 can also keep function operations not change this service outside this composite factor 240.Composite factor 240 restarts based on member in operation and reappraises.
For this application aims, telegon 250 indication framework resources are such as framework agency 266, for all resources in the framework are carried out Coordinating Activity.To describe in detail as following, these functions comprise that the startup member is to provide this service, make the member stop to provide this service, recover this composite factor 240, and task other member in this composite factor 240 who reapposes inefficacy member's specific composite factor.For example, when the member that service is provided for assembly was lost efficacy, Coordinating Activity just arranged (or a plurality of) other member to come to provide identical service for this assembly.
In Fig. 2, telegon 250 is described as to reside in and illustrates on the node that this " truly " composite factor 240 can be depending on and one of resident node of the member of composite factor 240 in cluster.The node of telegon 250 is considered to should composite factor 240 being thought resident node wherein by framework 260.As previously mentioned, to be considered to reside in this specific node wherein be optional to composite factor 240 and may change.Therefore, the position of this telegon 250 should not be considered to one than the prior node of other node.
Framework 260 thinks that the reason that composite factor 240 physics are present on certain node is that this framework self exists restriction.These restrictions in some cases, can prevent that framework 260 from making the function of these composite factor 240 execution separate with any specific node.Therefore, the embodiment that Fig. 2 describes has illustrated the telegon 250 that has the position on a node, has solved these possible restrictions of framework 260.If this composite factor is regarded as the specific node that resides in wherein by framework 260 and do not lost efficacy in particular instance, then the embodiment that describes of Fig. 2 can provide the service of this composite factor 240 effectively continuously.The generation if these lost efficacy, then this composite factor 240 will have to be resumed.This rejuvenation can comprise and restarts this composite factor 240, and this composite factor of reappraising is so that determine its each member's state.But in most of the cases, there are many nodes to provide the member for this composite factor 240.Because all nodes all can be counted as the resident position of this composite factor 240, therefore the thing that causes composite factor 240 to stop owing to a node failure is uncommon.
This composite factor 240 has solved the restriction of this framework 260, the performance and the availability that provide numerous characteristics to improve its member simultaneously.Such as, the member of operation can provide automatic framework operation and transparent load Sharing topology in composite factor 240.Other service that provides owing to composite factor 240 also is included as many clients that this composite factor can serve and provides and connect and the real-time working load management.This composite factor also is convenient to provide as required additional capacity.Such as, respond the specific predefine condition of the workload that this composite factor carries out, can increase the radix of composite factor 240 automatically.Can also can unification (commoditize) by the work that composite factor is carried out.
Telegon
As previously mentioned, the framework resource and the agency of telegon 250 representative execution Coordinating Activities.In one embodiment, telegon 250 also comprises the member's of this composite factor 240 resource, and their substitute framework resource or as these Coordinating Activities of additional execution of framework resource.In one embodiment, this telegon activity comprises each member who starts in this composite factor 240, stops each member, and recovers each member in this composite factor 240.The mode of recovering composite factor can be according to predefined agreement, such as the service of restarting this composite factor 240, and this composite factor member's that reappraises state.
In one embodiment, load management software arranges the special member of this composite factor 240 that the entity of service to this service of request is provided.When this member that this service is provided stops to provide this service (such as member or node failure), telegon 250 just arranges another member in this composite factor 240 to provide service to this entity.All active members in the composite factor are shared operating load in when operation, and are the assembly of the service that receives the self-disabling member this operating load of balance again when recovering and when repairing.
Whenever the member of this composite factor stops or losing efficacy, this composite factor is just reappraised.The member of composite factor 240 continues to be reappraised up to this composite factor in the operation of the outside of this composite factor.Be used to recover the function of this composite factor 240 by framework 260 and/or framework agency's 266 resource execution.The recovery of composite factor 240 is by separate member, specific framework resource and/or be not that other resource that is used for carrying out the resource collection of Coordinating Activity is carried out.
In one embodiment, telegon 250 is arranged service to offer and is resided in a request assembly in the node.In addition, this request assembly is corresponding to external modules such as for example client components.As long as this composite factor can be used, the Coordinating Activity that framework agency 266 carries out just keeps active state.Being used to provide the resource of telegon 250 also to be used for the keeper provides interface and composite factor 240 and member 214,224,234 thereof mutual.
As an example of system in operation, first assembly, 216 request composite factors 240 provide service.Asset creation from coordinated device 250 representatives of the request of first assembly 216.Telegon 250 arranges first member 214 that service is provided.Although can provide service to assembly 216, because first assembly 216 resides in first node 210, so telegon 250 can arrange first member 214 to provide service to first assembly 216 by any other member who forms composite factor.For service is provided, first member 214 can depend on the use of first resource 212.Such as, first resource, 212 corresponding interfaces or lower level are used, and are used for and other computing machine or database communication.
According to an embodiment, when the member of composite factor 240 stopped to provide service, system provided available service as described in Figure 1.If first member 214 was lost efficacy, another member of composite factor 240 can provide service to first member with comparable mode.First assembly 216 can switch to the service of reception from another member pellucidly, and does not significantly interrupt when receiving service.The quality of the service level that another member provides is in comparable or equal level.Therefore, if first member 214 was lost efficacy, then telegon 250 can arrange second member 224 to provide service to assembly 216 automatically.
Causing from the service failure that a member switches to another member may be because whole node lost efficacy.This inefficacy also may be that the member was lost efficacy, and node still keeps operable state.In addition, the resource of node may lose efficacy, thereby this composite factor 240 can not be operated.
In example as described in Figure 2, the failure type of demonstration is that first member 214 was lost efficacy, and the node 210 and first resource 212 still keep operability.As the response that detects inefficacy, telegon 250 arranges second member 224 on the Section Point 220 to provide this service to first assembly 216 automatically.This switching is transparent, and promptly Fu Wu recipient not can be appreciated that the place that service is provided has moved on to new position.Yet, may receive an interruption to the recipient of assembly.This interruption is in the lower level of network activity, thereby the effect of this interruption is transparent for the user.
In one embodiment, second member 224 uses second resource 222 that this service is provided, wherein second resource provide can with first resource, 212 comparable functions.In such embodiments, there is local dependence between the member 224 and second resource 222.But, when member 224 at a node and when another node provides the function of second resource 222, also may have overall dependence.
Management depends on the resource of service
In one embodiment, the service that is provided by composite factor 240 is provided first assembly 216.First assembly 216 is that " weak dependence " uses, and the service of its request composite factor 240 starts, but does not need the follow-up execution of this service.Therefore, when receiving the request that starts this first assembly 216, first assembly 216 can be to composite factor 240 request services.
As another example, the assembly of request service can be a client component, and it is positioned at outside the resident node of member.In this example, first assembly, 216 correspondences depend on by the application that can compare the service that resource 240 provides.The function using this service and provided by first assembly 216 can be provided client component.
Lost efficacy if the member occurs on first node 210, and first assembly 216 depends on the service that composite factor 240 provides, first assembly 216 can stop so, or inertia.Service can offer assembly by any other node of this service of supply.Such as, available from second member's 224 service, telegon 250 just arranges second member to provide service to the specific components that receives from first member's 214 service like this.Therefore, the service of composite factor 240 can offer assembly by second member 224 on the Section Point 220.In fact, in case first member 214 was lost efficacy, give other member as all task mean allocation that first member 214 of the part of composite factor 240 carries out.When second member 224 provided service to assembly, first resource 212 or second resource 222 can be used by second member 224.If client component needs second assembly 226, and second assembly 226 depends on this service, so just on second member 224, carry out second assembly 226 after this service of startup.
If first assembly, 216 weak dependence first members 214, if perhaps first assembly does not rely on first member 214, when first member was lost efficacy, first assembly 216 can keep operation on first node 210 so.First resource 212 does not rely on first member 214, so when the member was lost efficacy, it still kept operating with other application or the assembly of first node 210.
In inefficacy is under the situation of first node 210 complete failures, then shares in other active members that is operated in this composite factor 240 that first member 214 carries out.After first node lost efficacy, provide service to client component by making second member 224 on the Section Point 220, can continue to be provided for the service of client component.This second member 224 uses second resource 222 that service is provided.Second assembly 226 if desired, the work of service to assembly that provides of carrying out can also be provided for it.Other necessary resource starts on Section Point 220.The order that other resource starts is based on the dependence of setting up between the resource on the Section Point 220, comprises the dependence between other resource of composite factor 240 and Section Point.
Take place under the situation of inefficacy for the second time at the Section Point 220 or second member 224, then load management software arranges the work of second member execution to be assigned to other active members of this composite factor 240.In the example that Fig. 2 provides, in case first and second members were lost efficacy, all working that first member 214 and second member 224 carry out can be carried out by the 3rd member 234 of the 3rd node 230.In one embodiment, any one member in the composite factor 240 was lost efficacy and will be caused this composite factor to be reappraised.This is reappraised and can be carried out by the resource of telegon 250 representatives, or is carried out by other resource on the framework 260, or is carried out by the member of this composite factor 240.When composite factor 240 is done as a whole whole inefficacy, just start resuming work of composite factor 240.In one embodiment, this does not have other movable member that the situation of service is provided corresponding to working as.In another embodiment, for this composite factor is specified minimum cardinality, and as long as the quantity of this active members satisfies this radix, this composite factor just is considered to move.
In one embodiment, the recovery that starts composite factor 240 need verify that each member 214,224,234 stops.Each member 214,224,234 who starts this composite factor of the recovery pattern of wants of composite factor 240 starts.In addition, the number of members that has started satisfies the specific radix that the recovery needs that start composite factor 240 are verified this composite factor 240.The member who is started can start voluntarily, or starts by a certain activity of other resource relevant with this composite factor 240.According to an embodiment, when one or more members were lost efficacy and/or recover, this member just recovered independently, this composite factor of just reappraising.This composite factor needn't recover (unless the physical node that it moves lost efficacy)
Virtual composite factor
Fig. 3 has described another embodiment, and wherein system 300 is configured to comprise the virtual composite factor that service is provided.Be similar to the described true composite factor of Fig. 2, described wherein virtual composite factor 340 and provide the system of service for residing in node cluster inside or outside entity.A feature of virtual composite factor is that it does not all have physical interdependence to any one node.On the contrary, virtual composite factor form with logical circuit or software on each node is evenly distributed its process, and has other assembly of physical location to carry out any processing on certain contact without framework or this composite factor.This process is distributed, in case so that any one node failure in the cluster, this process still keeps and can carry out.Therefore, use the embodiment of virtual composite factor to solve the physical restriction of true composite factor, and same basic function and characteristic is provided simultaneously.
In one embodiment, this system comprises first node 310, Section Point 320 and the 3rd node 330 (representing n node).First node 310 comprises first resource 312, first member 314 and first assembly 316.Section Point 320 comprises second resource 322, second member 324 and second assembly 326.Similarly, the 3rd node 330 comprises information resources 332, the 3rd member 334 and the 3rd assembly 336.First member 314, second member 324 and the 3rd member 334 form virtual composite factor 340.
In framework agency 366 each that reside in first node 310, Section Point 320 and the 3rd node 330.Framework agency 366 communicates by letter with framework 360.In framework 360, be provided with register 362.Register 362 is addressable for framework agency 366.Framework agency 366 allows to regard framework 360 as across the resident node of the member of composite factor 340 unity logic entity.
Virtual compound
In Fig. 3, (virtual composite manager, VC) logic of Coordinating Activity is carried out in 350 expressions with virtual Composite manager.VC 350 resides on the framework 360.This VC 350 just carries out the representative of process, resource and the logic of management and/or Coordinating Activity for composite factor 340.Shown VC 350 resides on the framework 360, but the code that is used for carrying out VC might reside in node.In one embodiment, VC 350 carries out on framework 360, or carries out by framework agency 362, or carries out on all nodes comparably.In addition, virtual composite factor 340 is not regarded as by framework 360 and is positioned at any one node, and therefore virtual composite factor 340 can be independent of these nodes and operates.But this VC350 in fact comprises the code of being carried out by framework 360, framework agency 366 and/or composite factor member.A node failure in the cluster will can not influence the activity and the function of the resource execution that comprises VC 350.
The function that comprises the resource execution of VC 350 comprises auxiliary or management member function.Member's function comprises the member who restarts after the inefficacy.VC 350 is also according to its member's the state composite factor of reappraising.This is reappraised and can respond member and lost efficacy and finish.VC 350 can trigger the recovery of composite factor 340 when the member of this composite factor 340 is lost efficacy.One or more other members that work that VC 350 can also carry out member or task are repositioned onto this composite factor.
In one embodiment, VC 350 run time versions collect relevant member 314,324 and 334 with and the status information of corresponding node 310,320,330.VC 350 also collects the status information when changing in member 314,324 and 334.Whether this status information indicates any member 314,324 and 334 movable, and perhaps whether certain specific member was because inefficacy stopped to provide service.This status information can be collected in and be arranged in framework 360 inner or outside registers 362.Because VC 350 can be carried out on framework 360 by all nodes, so no matter this node failure is owing to node itself or member, the influence that this VC was not lost efficacy by certain special member in the composite factor 340.
Composite factor 340 can be configured to use in system by user's input.Keeper or other user's input can be specified the configuration information of composite factor 340.When framework 360 comprises being used to of being pre-existing in when setting up this composite factor 340 and carrying out the process of the VC that is used for this composite factor, configuration data is used to state composite factor 340.Therefore, in one embodiment, by the input configuration data, this keeper can develop and manage composite factor 340, and any code need not be provided.
The class of operation of composite factor 340 and VC 350 is similar to the operation of the described true composite factor 240 of Fig. 2.Difference is, in Fig. 2, if the resident node failure of telegon 250, then composite factor 240 is disabled.In the described system of Fig. 3, any one node failure, even some node failures in this cluster can not forbidden this composite factor 340.As a result, this virtual composite factor 340 is more reliable.
When the request made the service of composite factor 340, VC 350 arranges among the members 314,324,334 to provide service to the request assembly.Therefore, as shown in Figure 3, as the response to 316 these services of request of first assembly, VC 350 arranges first member 314 that service is provided.When first member 314 was lost efficacy, VC 350 arranged second member 324 that this service is provided.If of depending in the resource 312,322,332 of this composite factor 340 provides service, VC 350 will provide second member to use first resource 312 or second resource 322 so.
Under the situation that node failure takes place, represent the VC 350 of resource to arrange second member 324 on the Section Point 320 that service is provided.If node or member were lost efficacy, then for example the client waits the service of external module reception from second member 324.For example, if first assembly 316 depends on composite factor 340 and first member 314 was lost efficacy, then can provide service to first assembly by second member 324.In addition, the function that is provided by first assembly 316 on the first node 310 can be substituted by second assembly 326 on the Section Point 320.
Composite factor 340 model simplifications easy to use constitute each member's of composite factor configuration and use.More specifically, the distributed logic notion easy to use that is provided by composite factor 340 " anonymous (anonymity) " and " radix (cardinality) " use the user and dispose this composite factor." anonymity " and " radix " will be explained in more detail below.Although the notion of " anonymity " and " radix " is to describe in the context of virtual composite factor 340 (background) environment, these notions also can realize on true composite factor as shown in Figure 2.But these notions are more convenient for combining with virtual composite factor 340, do not coordinate service between the member because the latter's member relies on any specific node, are therefore providing aspect the availability all nodes to be treated coequally.The member who operates on these nodes only need discern by the relation of itself and composite factor 340, and the particular address or the sign of the resident node of this member are discerned.Anonymous and radix allows user or keeper that composite factor is done as a whole addressing, and member that need not be independent sign.
Anonymous
According to an embodiment, the assembly of the service by allow using composite factor (no matter being " really " or " virtual ") uses those resources and need not know any separate member trimscript will acquisition anonymity really in composite factor.Anonymity makes user or system manager more easily manage member in the composite factor 340.Because member 314,324,334 need not have unique name word mark (at least for user or keeper), so anonymous further enhancing in virtual composite model.Equally, member 314,324,334 does not need independent addressable yet.Needn't be such as, keeper separately to each node addressing when attempting to realize certain customized configuration of composite factor 340.
According to an embodiment, user or keeper are according to the member of the virtual composite factor 340 of relation recognition of they and composite factor.Such as, specific service can be specifically specified in the instruction of configuration composite factor 340, and this is enough for providing the concrete member of this service to certain with this application of instruction.This forms contrast with mode of using unique identifier and address to manage each application (member who comprises composite factor 340) more traditionally.
Can be by will for example using or resource such as example is claimed as " this locality " resource and obtains anonymity." this locality " resource can belong to a set, and the identifier of the node that name that this resource can be gathered and local resource are resident identifies.Such as, a plurality of examples of one group of corresponding application of resource, and this example can be identified by apply names.
In having a plurality of members' system, because the keeper need manage the member by the identifier and the address of specifying each member, the burdensome so Admin Administration member's task becomes.But anonymity makes the keeper manage application by specifying composite factor (rather than its independent member) and the resident node of interested member.
In one embodiment, framework 360 can be configured to comprise mapping logic, shines upon unique node label and the address information node address to composite factor 340 and this composite factor of support.This has simplified the member's of Admin Administration's composite factor 340 mode.
Therefore, in system as shown in Figure 3, can define local resource by the unique identification symbol, and on some or all nodes that constitute composite factor 340, carry out the transplanting of cluster wide.In this way, by discerning this composite factor 340 and the resident node of this member, the member can be by operation separately, configuration or management.This scheme forms contrast with the present scheme of management node cluster, each resource of the latter by its own unique identifier with and the identifier and/or the address of node and its architecture identified.
Radix
According to an embodiment, radix is meant the number of active members in the composite factor.When the special member of composite factor was more important unlike other member, the service level of specifying this composite factor to provide was provided the radix of composite factor.
In another embodiment, radix is detected by some performance criteria.The time of the service that can corresponding virtual composite factor 340 provides such as, radix.
In one embodiment, radix is used for determine whether the lowest service level is provided by composite factor.Can specify as long as the quantity of active members is equal to or greater than " 1 " or a certain other numeral such as, keeper, then this composite factor just is considered to movable.
In another embodiment, during the configuration of composite factor 340, use radix.Such as, the keeper can go up the configuration information of realizing and dispose this composite factor 340 by the member who specifies in certain specified quantity.
As another example, radix can be used to expand the service that composite factor 340 provides.Can increase the radix of appointment, so that increase the work that composite factor 340 is carried out.The keeper can use the radix of this appointment, and need not specify any relevant special member or the information of the node of service is provided.Therefore, the cardinal sum anonymity can be mutually promoted.
Based on service level management composite factor
Fig. 4 has described the method based on a plurality of assemblies of service level management composite factor.Method can be as Fig. 2 or realize in the system as shown in Figure 3 as described in Figure 4.For purposes of illustration, when describing the step of this method, referring to the element of Fig. 3.
In step 410, use the member 314,324 and 334 who resides in node 310,320 and 330 respectively to set up composite factor 340.Each member who forms composite factor 340 can both provide comparable service.This composite factor 340 can be a homogeneity, and promptly each member 314,324 and 334 provides the service that is equal to fully.In addition, this composite factor 340 can be heterogeneous, or has corresponding to service quality difference or the different different attribute of execution service manner.
In step 420, by the data configuration composite factor 340 of keeper's input.In one embodiment, this configuration can be specified the service level of composite factor 340.Particularly, configuration data can be specified the radix of composite factor 340, by the integrity service level of measuring service time, or the integrity service level of measuring by handling capacity, or some further features.In addition, configuration data can be specified about this composite factor 340 and member 314,324 and 334 out of Memory of how operating or recovering.The dependence (weak or strong) that can specify the hypothesis about the required assembly of the required resource of this composite factor 340 and this composite factor 340 to exist such as, this configuration data.
Step 430 provides each member 314,324 and 334 collections of status information from composite factor 340.This step can be performed by the verification operation of carrying out composite factor 340.Status information can be collected by logic element that forms VC 350 or process.Such as, each node comprises the logic as VC 350 parts.This logic is collected the status information of the member on relevant certain specific node.This status information is stored in the register 362.In one embodiment, corresponding certain special member of status information is movable or shut-down operation.Certain special member change state is responded, collect this status information.
In step 440, determine the service level of composite factor 340 and certain specified level corresponding to inactive composite factor 340 are compared.In one embodiment, specify concrete level by the keeper.Such as, this service level is corresponding to the radix of user's appointment, and wherein exercisable number of members must equal or exceed this radix.As another example, this service level is to certain number percent or the ratio of the maximum service quality that should composite factor 340 can carry out.
If determine that in step 440 this composite factor 340 is available, so in step 450, one or more members that VC 350 is arranged in the composite factor 340 provide service to requesting component.This service be provided to always to the service level that should serve be confirmed to be unreachable till.As long as this service surpasses specified level, VC 350 just guarantee member 314,324 and 334 at least one providing the service of composite factor 340 to requesting component.If be arranged to provide the member of service to lose efficacy, VC 350 provides service to requesting component with regard to arranging another movable member so.As previously mentioned, member's transfer can seamlessly be carried out, and the switching between the member is not measurable (determinable) for requesting component like this.
If determine that in step 440 this composite factor 340 is unavailable, composite factor 240 is with regard to off-line so.The member can restart, and this causes this composite factor 340 to be reappraised.Therefore, composite factor 340 function that will exist with its member and existing.
In addition, composite factor 340 is expanded so that the Additional Services that this composite factor 340 can be used to be provided.Such as, the radix of composite factor can be expanded.Can set up the inertia member such as, composite factor and gather secondary node as the active members of this composite factor 340.Under the situation of service of expansion composite factor 340, then the resource of VC 350 activates one or more members from standby or inactive member.
Monitor interested incident
Fig. 5 shows and is configured to detect immediately and the system of active with interested framework event communication.System as shown in Figure 5 can realize with the system of Fig. 2 and Fig. 3.Usually describe with reference to the unit of Fig. 3 and to be implemented in example components embodiment illustrated in fig. 5.
System 500 is included in the application example of carrying out on the node 530 510.This node 530 comprises framework agency 540, and the latter communicates by letter with the framework that the node cluster is shared.According to an embodiment, watch-dog 520 is associated with application example 510, to detect interested incident immediately after incident takes place.When interested incident 524 took place, watch-dog 520 transmitted event notification 522.Framework agency 540 carries out polling operation 526 according to the predetermined time interval to using example 510, so that determine the state of application example.Interested incident 524 can change the state of application example 510, acts on behalf of 540 by the identical information that poll obtains to framework so that the generation of this incident can make watch-dog 520 provide with it.Therefore, polling operation 526 and watch-dog 520 are carried out overlapping function, and this redundancy guarantees that system 500 is reliable.Can know that in the operating process of node 530 watch-dog 520 is that poll can be alternative fully by the watch-dog of event system and application level in this case reliably for using example 510.In addition, poll and monitoring can be used together to detect interested incident fast and reliably.
In one embodiment, interested incident cover framework 540 employed states.Therefore, incident is corresponding with following some: (i) when application example 510 startups; (ii) when application example 510 is repositioned onto another node; (iii) when but application example 510 stops not lose efficacy; (iv) when application example 510 stops and losing efficacy; And (v) when with the application example inspection for can operate the time.Detect these incidents one of them the time just start (initiate) framework agency 362 and carry out some activity.This activity also triggers VC 350 and carries out one of its activity, such as beginning to check or begin recovery.
In configuration as shown in Figure 3, for each member 314,324 and 334 sets up watch-dog 520.From the request of another assembly, watch-dog 520 detects the service when first members 314 stop maybe will stopping to provide composite factor 340 such as, response.Event notification 522 sends from watch-dog 520 and lost efficacy to framework 360.Respond this communication and be provided, the VC 350 that resides on the framework 360 arranges second member 324 in the composite factor 340 that this service is provided.
Some or all status information can obtain by the poll 526 of framework 540.May be always not reliable for watch-dog 520.Such as, watch-dog 520 can lose efficacy with member 510 or this node.In this case, the framework 540 of execution poll can provide reliable backup.
The enables/disables feature
The member of composite factor may suffer inside the plan or planless inefficacy.Can the inefficacy in the executive plan for maintenance.Planless inefficacy occurred in the member when losing efficacy, or it is when depending on another failed resource.In this case, be this member's extraction system useful.In as Fig. 2 and 3 described systems, the control of this member being taken out framework is useful.This means when this member is lost efficacy and to restart automatically.On the contrary, this member should keep forbidding up to taking compensation (corrective) activity to repair member or the resource that is relied on.
Fig. 6 has described the basic fundamental of management composite factor when the member was lost efficacy.Fig. 7 has described another technology of managing composite factor after node is disabled.In the description that is provided, the reference purpose of label shown in Figure 3 is to explain the example components of using with described technology.
At Fig. 6, step 610 provides by a plurality of members 314,324 and 334 on node 310,320,330 clusters that constituted and has set up composite factor.For illustrative purposes, suppose that first member 314 provides this service at first.
In step 620, determine that first member 314 has stopped providing service.For example, consult to make that this is definite by watch-dog 520 or by poll and first member 510 by framework agency 540.
Step 630 provides in order to carry out maintenance function on this member and has forbidden this member.This maintenance function can be manual or automatic.If first node 310 is disabled, first member 314 just can not restart.First member 314 keeps forbidding and inertia till being enabled afterwards always.In other words, the forbidding to first member 314 continues.Therefore, if the member is disabled and first node is unloaded, restarting first node so will can not make the member 314 that wins restart automatically.
It is disabled that member or its node can be used as the part of inside the plan maintenance.Therefore, the member was lost efficacy, and must not to trigger node disabled.In one embodiment, inefficacy does not cause the member disabled.On the contrary, forbidding is inside the plan operation, is used for expansion reparation or schedules operations on the node.
In the method for Fig. 7, step 710 provides the request of reception to restart first member 314.This request is received after first member 314 has stopped carrying out the service of composite factor 340.
In step 720, determine whether certain specific member is enabled or disabled.This state can extract (retrieve) from register 362, perhaps extract from other memory module relevant with framework 360.Therefore, register 362 can write down member 314 and enables or forbid before the member just stops service being provided.
If determine that in step 720 first member 314 is disabled, so in step 730, first member 314 does not restart.When this occurs in first member 314 and carries out inside the plan the maintenance.In one embodiment, this another member that can trigger in the VC 350 arrangement composite factors 340 provides service to requesting component.
If determine that in step 720 first member 314 is enabled, just attempt restarting automatically first member 314 in step 740 so.If first member can be restarted, first member just adds composite factor 340 again so.
Describe as Fig. 7, the member's of composite factor enable/disable state is configured to continue, and determines whether to attempt restarting this member.This member can be disabled, so that it can not restarted in start-up course automatically.
Forbidding/enabled state can expand to composite factor.Composite factor can be used as whole disabled, so that all members are disabled.In addition, composite factor can be enabled, and this will cause all not have disabled member to be enabled.Therefore, the independent member of forbidding will cover enabling composite factor.As a result, need not forbid the member that whole composite factor just can be managed inefficacy.
Ardware overview
The block scheme of Fig. 8 has shown the computer system 800 that realizes embodiments of the invention.Computer system 800 comprises that bus 802 or other carry out the communication mechanism of information communication, and is coupled to the processor 804 that bus 802 is carried out information processing.Computer system 800 also comprises primary memory 806, such as random access storage device (RAM) or other dynamic storage device, is connected to bus 802, is used for the instruction that canned data and treating carried out by processor 804.Primary memory 806 can also be used for storage temporary variable or other intermediate information between the order period that execution is treated to be carried out by processor 804.Computer system 800 further comprises ROM (read-only memory) (ROM) 808 or is coupled to bus 802 and is used for storing the static information of processor 804 and other static memory of instruction.Memory device 810 (such as disk or CD etc.) is provided and is coupled to bus 802, is used for canned data and instruction.
The present invention relates to be used to the use of the computer system 800 of the technology implementing to describe herein.According to one embodiment of present invention, answer processor 804 is carried out the one or more sequences that are included in the one or more instructions in the primary memory 806, and these technology are carried out by computer system 800.These instructions are read in primary memory 806 from for example another computer-readable medium of memory device 810 grades.The instruction sequence that execution is included in the primary memory 806 will cause processor 804 to carry out the process steps that describes below.In another embodiment, hard-wired circuit is used to substitute or realize invention in conjunction with software instruction.Therefore, realization of the present invention is not limited only to the particular combinations of any hardware circuit and software.
The term of Shi Yonging " computer-readable medium " refers to participate in providing instructions to any medium that processor 804 is carried out herein.This medium has many forms, includes but not limited to non-volatile media, Volatile media and transmission medium.For instance, non-volatile media comprises, CD or disk, for example memory device 810 etc.Volatile media comprises dynamic storage, such as primary memory 806.Transmission medium comprises concentric cable, copper conductor and optical fiber, comprises the cable of forming bus 802.Transmission medium also comprises sound or form of light waves, such as the ripple that produces in radiowave or infrared data communication.
For instance, the common version of computer-readable medium comprises: floppy disk, flexible disk, hard disk, tape, or any other magnetic medium, CD-ROM or any other light medium, card punch, paper tape, any other has the physical medium of sectional hole patterns, RAM, PROM, and EPROM, FLASH-EPROM, any other storage chip or tape, the carrier wave that describes below, or any other computer-readable medium.
Various forms of computer-readable mediums can be used for one or more sequences with one or more instructions sends processor 804 to and is used for carrying out.For example, this instruction can be carried on the disk of remote computer at first.This remote computer is written into its dynamic storage with instruction, and uses modulator-demodular unit to send instruction in phone line.The local modem of computer system 800 can receive the data on the telephone wire, and uses infrared transmitter that data are converted to infrared signal.Infrared detector can be received in data carried by data in the infrared signal, and suitable circuit is placed into these data on the bus 802.Bus 802 carrying data are to primary memory 806, and processor 804 is retrieval and execution command from primary memory 806 again.Before or after processor 804 was carried out, the instruction that primary memory 806 receives was stored on the memory device 810 alternatively.
Network connects 820 provides data communication to arrive other data set by one or more networks usually.Connecting 820 such as, network can provide by local network 822 to principal computer 824 or by the connection of the data set of ISP (ISP) 826 operations.ISP 826 provides data communication services by the ten thousand dimension packet data communication network that are commonly referred to " internet " 828 again.Local network 822 and internet 828 all use electric signal, electromagnetic signal or light signal carrying digital data stream.Through the signal of diverse network, the signal on the network link 820 and through the signal of communication interface 818, carrying and delivering to and, be the demonstration form of the carrier wave of transmission information from the numerical data of computer system 800.
After receiving code, the code that receives can be carried out by processor 804, or is stored in memory storage 810 or other nonvolatile memories and is used for carrying out subsequently.By this way, computer system 800 can obtain the application code of carrier format.
In above-mentioned instructions, the present invention has been described with reference to specific embodiment.Yet, obviously can make various variations and change, and not deviate from the spirit and scope of broadness of the present invention it.Therefore, instructions and accompanying drawing only are used for illustrating and non-limiting the present invention.
Claims (30)
1. method of managing a plurality of assemblies from framework said method comprising the steps of:
Foundation comprises a plurality of composite factors that reside in a plurality of members on a plurality of nodes in the network work system, and each member among wherein said a plurality of members can both provide utility, as long as wherein satisfy specified conditions, and described composite factor energy maintenance activity;
Use each member's of the described composite factor of framework monitoring resource state;
Making needs the assembly of described utility from the described utility of described framework resource request;
Provide described utility to described assembly by following steps:
Arrange a member in the described composite factor to provide described utility to described assembly by described framework resource, the arbitrary active members in the wherein said composite factor can be used to provide described utility to described assembly;
When the described member who is arranged to provide described utility stops action, make described utility offer described assembly automatically by another member among described a plurality of members, wherein arbitrary active members can be used to provide described utility to described assembly; And
Be independent of each member's of described composite factor state, safeguard the state of described composite factor, make that the described state of described composite factor is with regard to the maintenance activity as long as described specified conditions are met.
2. method according to claim 1, wherein said a plurality of member comprises active members set and inertia member set, and wherein at least one member in described active members set provides described utility according to the mode that is different from the employed mode of at least one member in described active members set and described inertia member set.
3. method according to claim 1 further comprises: the one or more members among described a plurality of members are stopped action make response, to assess the step of described composite factor state.
4. method according to claim 1 further comprises the step that disposes described composite factor according to user-defined standard by the framework resource.
5. method according to claim 1 further comprises: use described framework resource to stop described composite factor, so that the step that each member among described a plurality of member stops.
6. method according to claim 2, further comprise the source of described composite factor never the resident node of member in the active members set be repositioned onto the resident node of member in the active members set, wherein just before the step in source of described composite factor is reorientated in execution, at least one of described a plurality of active members is the member in described inertia member's set.
7. method according to claim 1 further comprises from the performance characteristic of the described composite factor of described framework resource measurement.
8. method according to claim 1 further comprises interface is provided, and makes the keeper carry out one or more operations to described composite factor by described framework resource.
9. method according to claim 1 further comprises making described assembly can depend on described utility, and does not consider that among described a plurality of member which providing described utility.
10. method according to claim 1, each that set up wherein that composite factor comprises described a plurality of members of making in the described composite factor can both be used with the resource collection on the described member node, and wherein said member requires the resource collection on the described member node that described utility is provided.
11. method according to claim 1 wherein provides described utility to comprise to the step of described assembly:
First member who arranges described composite factor provides described utility to described assembly from described first member's node;
When described first member was lost efficacy, making provided described utility to described assembly by second member among described a plurality of members from described second member's node, and wherein said second member's node is different from described first member's node; And
Wherein before described first member was lost efficacy, described method further comprises makes described second member can use the resource collection on described second member's node that the step of described utility to described assembly is provided, and described second member requires the described resource collection on described second member's the node that described utility is provided.
12. method according to claim 1 wherein provides described utility to comprise that to described assembly execution is from by the data source reading of data of described a plurality of nodes sharing with write data at least a the operation of described data source.
13. method according to claim 11, comprise further whether detect described first member lost efficacy, unless described first member's node still can be operated, wherein carry out making described second member that the step of described utility to described assembly is provided, although one or more other resources are exercisable on described first member's the node.
14. method according to claim 13 further is included in the step that detects described first member of resetting automatically after described first member was lost efficacy.
15. method according to claim 13 wherein when attempting recovering described first member, carries out making second member that the step of described utility to described assembly is provided automatically.
16. method according to claim 1 is wherein set up composite factor and is comprised a plurality of examples of database application are carried out on a plurality of nodes.
17. method according to claim 1, wherein said framework carry out with described composite factor regard as reside in a specific node in described a plurality of node to small part on.
18. method according to claim 1, wherein said framework are assigned on described a plurality of node.
19. method according to claim 1 further is included in and carries out the step of using on the first node of described a plurality of nodes, wherein said application-dependent is in described composite factor.
20. method according to claim 19, comprise that further first member from described a plurality of members provides described utility to the described application of carrying out at described first node, wherein said first member resides on the described first node, wherein stop to provide described utility in response to described first member, described method further comprises the execution that stops described application.
21. method according to claim 19, comprise that further first member from described a plurality of members provides described utility to the described application of carrying out at described first node, wherein said first member resides in described first node, wherein stop to provide described utility in response to described first member, second member that described method further comprises among described a plurality of members on the Section Point from reside in a plurality of nodes provides utility to the application of carrying out at described first node.
22. method according to claim 1, wherein said method further comprises:
When described composite factor is moving, start the part of member as described composite factor; And
Make response to starting described member as the member of described composite factor, make the described member who is activated guarantee that described composite factor moves.
23. method according to claim 1, wherein said specified conditions are corresponding to the active members of the minimum number of the user's appointment in described composite factor.
24. method according to claim 1, wherein said specified conditions are corresponding to the minimum service level of the user's appointment that is used for described composite factor.
25. a method that is used for managing from framework a plurality of assemblies, described method comprises:
Foundation comprises more than first member and more than second member's composite factor, described more than first member and more than second member reside on a plurality of nodes in the network work system, among described more than first member each all is movable, in order to utility to be provided, each among described more than second member can provide described utility when being activated;
Arrange the member among described more than first member that first assembly of described utility to the described utility of request is provided by described framework, wherein, arbitrary member among described more than first member all can be used to provide described utility to described first assembly, in response to the inertia that becomes of the one or more members among described more than first member, carry out following steps:
The member of activation in described more than second member; And use the active members among described more than first member or described more than second member that described utility is provided automatically.
26. method according to claim 25, the state that wherein is independent of in described composite factor each movable member is kept the state of described composite factor, as long as make and satisfy certain conditions, the state of described composite factor is with regard to the maintenance activity.
27. method according to claim 25, wherein use the active members in described more than first member or more than second member to provide described utility to comprise the radix of keeping described member automatically, wherein said member provides the described utility of described composite factor versatilely.
28. method according to claim 27, thereby comprise that further a plurality of members by activating among described more than second member provide described utility to expand the step of described radix.
29. method according to claim 25 wherein uses the resource of described framework to arrange the member among described more than first member that described utility is provided, described framework is shared by described more than first member and described more than second member.
30. method according to claim 29 further comprises by described framework resource and keeps status information about described more than first member and described more than second member.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US33434901P | 2001-11-30 | 2001-11-30 | |
| US60/334,349 | 2001-11-30 | ||
| US60/426,587 | 2002-11-15 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1723441A CN1723441A (en) | 2006-01-18 |
| CN100397349C true CN100397349C (en) | 2008-06-25 |
Family
ID=35912866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB028234367A Expired - Lifetime CN100397349C (en) | 2001-11-30 | 2002-12-02 | Real composite objects for providing high availability of resources on networked systems |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100397349C (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2360866A1 (en) * | 2010-02-12 | 2011-08-24 | Panasonic Corporation | Component carrier activation and deactivation using resource assignments |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5463733A (en) * | 1993-06-14 | 1995-10-31 | International Business Machines Corporation | Failure recovery apparatus and method for distributed processing shared resource control |
| US6134673A (en) * | 1997-05-13 | 2000-10-17 | Micron Electronics, Inc. | Method for clustering software applications |
| CN1312922A (en) * | 1998-08-11 | 2001-09-12 | 艾利森电话股份有限公司 | Fault tolerant computer system |
-
2002
- 2002-12-02 CN CNB028234367A patent/CN100397349C/en not_active Expired - Lifetime
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5463733A (en) * | 1993-06-14 | 1995-10-31 | International Business Machines Corporation | Failure recovery apparatus and method for distributed processing shared resource control |
| US6134673A (en) * | 1997-05-13 | 2000-10-17 | Micron Electronics, Inc. | Method for clustering software applications |
| CN1312922A (en) * | 1998-08-11 | 2001-09-12 | 艾利森电话股份有限公司 | Fault tolerant computer system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1723441A (en) | 2006-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6145089A (en) | Server fail-over system | |
| CN100549960C (en) | The troop method and system of the quick application notification that changes in the computing system | |
| US11516072B2 (en) | Hybrid cluster recovery techniques | |
| CN102656565B (en) | Failover and recovery for replicated data instances | |
| JP4721195B2 (en) | Method for managing remotely accessible resources in a multi-node distributed data processing system | |
| US7451215B2 (en) | OPC server redirection manager | |
| JP4637842B2 (en) | Fast application notification in clustered computing systems | |
| US7475127B2 (en) | Real composite objects for providing high availability of resources on networked systems | |
| US6990606B2 (en) | Cascading failover of a data management application for shared disk file systems in loosely coupled node clusters | |
| CN101137984B (en) | Systems, methods, and software for distributed loading of databases | |
| US20020152423A1 (en) | Persistent session and data in transparently distributed objects | |
| US20170315886A1 (en) | Locality based quorum eligibility | |
| CN102640108A (en) | Monitoring of replicated data instances | |
| CN110046039A (en) | Method for scheduling task and system, control centre's server and storage medium | |
| US6587860B1 (en) | Apparatus and method for tracking access to data resources in a cluster environment | |
| JPH10214199A (en) | Process restarting method, and system for realizing process restart | |
| CN105511987A (en) | Distributed task management system with high consistency and availability | |
| CN110417600B (en) | Node switching method and device of distributed system and computer storage medium | |
| US20050234919A1 (en) | Cluster system and an error recovery method thereof | |
| CN110798339A (en) | Task disaster tolerance method based on distributed task scheduling framework | |
| CN113377535A (en) | Distributed timing task allocation method, device, equipment and readable storage medium | |
| CN100397349C (en) | Real composite objects for providing high availability of resources on networked systems | |
| CN100563233C (en) | Fault-tolerance method during a kind of Common Object Request Broker Architecture is used | |
| CN103780433B (en) | Self-healing type virtual resource configuration management data architecture | |
| CN114201473A (en) | Log management method and server cluster |
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 | ||
| CX01 | Expiry of patent term |
Granted publication date: 20080625 |
|
| CX01 | Expiry of patent term |