Patent Application
20110191460
The present invention relates to networks supporting group communication and group management. It particularly relates to a node, a system and a method for group management.
Network management is becoming a more and more challenging issue due to the fact that networks become more and more complex, they grow in a forseeable way as well as in unforseeable ways, evolve continuously and require interoperability with other networks. An increasingly complex and heterogeneous network environment is also expected due to the emergence of an enormous amount of new, complex and advanced services. Different standards are implemented, management and billing service provisioning should be provided in a satisfactory manner both for customers, network operators and management staff. Over the last years, group communication has become an interesting issue. Numerous grouping strategies targeting distinct applications and designed to operate in specific networks have been proposed. A plurality of grouping protocols are known, such as for example the traditional token-ring protocols, multicast protocols and application-level and overlay-based grouping protocols.
By means of token ring protocols, a ring-based topology is created which is used to pass a token around the group members for e.g. controlling access to a given resource. Such protocols typically target small size networks. So called multicast protocols, on the other hand, focus on large scale networks. IP (Internet Protocol) multicast allows data to be sent to multiple receivers independently of their localization. The sources only need to know a multicast address and do not process any membership related information. As a consequence thereof, it becomes possible for members to join or leave a group easily since there is no need to register with a centralised group management entity. However, IP multicast has not been deployed to a very large extent both for technical reasons, such as complex billing, management and security concerns, but also for non-technical reasons. Therefore application-level based on overlay-based protocols have been suggested which have been more popular. Application-level grouping is for example discussed in “Void: Your Own Internet Distribution”, Paul Francis et al., http://www.isi.edu/div7/yoid/, and the NICE protocol “Scalable Application Layer Multicast” S. Banerjee, B. Bhattacharjee, C. Kommareddy, Proceedings of ACM Sigcomm 2002, Pittsburgh, Pa., August 2002. It is based on end hosts implementing group-related functionalities such as membership management. The underlying physical topology is hidden and a virtual graph is created between all the nodes. The overlay based approaches are for example described in “Scribe: A large scale and decentralized application-level multicast infrastructure” by M. Castro, et al. in IEEE Journal on selected areas in communications, Vol. 20, No 8, 2002.
Such overlay based approaches first build a DHT-based overlay assigning some coordinate or identity to each member so as to create a well connected virtual network, corresponding to an overlay network, on top of which a group is created.
The fact that grouping management is a fundamental concept in computer science, has had as an effect that a large variety of protocols have been developed, each creating different grouping topologies adapted to application requirements, for example bandwidth availability for real-time video delivery or video conference. However, a group management strategy, which includes a membership strategy and a group structure management strategy, raises a number of issues which cannot be satisfactorily handled by the current grouping protocols. Indeed, it typically requires human intervention to set up complex network element grouping in a large-scale network. The routing protocol OSPF (Open Shortest Path First), cf. OSPF version 2, IETF RFC (Request for Comments) 1247, www.ietf.org., July 1991 by J. Moy, requires the definition of separate areas in which traffic induced by flooding of link state information can be confined. However, this area definition still has to be performed statically by a network administrator. In addition thereto, grouping protocols only verify one or a limited set of constraints, and use one or a limited set of metrics to guarantee their performances. For example, a metric can impact on the structure of the group, elements can be grouped in an ordered way according to their bandwidth, or their distance (geographical or number of hops). Constraints may relate to the maximum number of elements that are grouped, minimum capacity required to group the network elements, for example processing power, memory, storage space, or the role of the network elements, (for example Radio Base Stations (RBS) or Radio Network Controllers (RNC) or any other nodes or network elements). This simplicity of the grouping protocols remains a major obstacle for configuring and managing dynamically sets of networked resources and network elements that are heterogeneous in nature, complex to manage, and dynamically evolved.
Thus, so far no satisfactory way of providing for grouping or grouping management has been provided.
It is an object of the present invention to provide means for network management implementing grouping. It is also an object of the invention to provide a solution that enables efficient implementation of group communications and group management, and particularly to provide a solution which enables automation of network management as well as increased performance thereof. Still further is an object of the invention to provide a solution as far as network management is concerned through which operation and maintenance traffic can be reduced and through which network management efficiency in general can be increased. Particularly it is an object of the invention to provide a node through which on or more of the above mentioned objects can be achieved, as well as a system and a method respectively, through which one or more of the above mentioned objects can be achieved.
Therefore the present invention suggests a node supporting networking and communication with other nodes by means of a layered communication protocol comprising a communication layer and which further supports group communication with other nodes controlled via a grouping protocol which is adapted to interact with said communication layer. According to the invention the node comprises a policy engine which is arranged separately from, but in communication with, said communication protocol. The policy engine is adapted to establish current and/or changing node, group and network related information and provide or activate one or more grouping policies and to, using said policies and said information, dynamically and automatically control grouping related actions and adapt the grouping.
The invention also suggests a data or telecommunications network comprising a number of nodes having the features described above.
In order to solve one or more of the problems discussed above, i.e. to fulfil one or more of the objects referred to above, the invention also suggests a method of controlling grouping of nodes supporting communication and networking with other nodes by means of a layered communication protocol comprising a communication layer, and a grouping protocol. According to the invention the method comprises the steps of; providing or activating one or more grouping related policies to be used for grouping of nodes; monitoring or establishing updated, current (or changing) node, group and network related information; using said information and grouping policy or policies to control grouping related actions dynamically and automatically and to adapt the grouping.
It is an advantage of the invention that self-grouping of network elements or nodes can be provided. Generally it is an advantage that network management performance can be increased and made more efficient and that operation and maintenance related traffic significantly can be reduced. Still further it is an advantage that human interaction or involvement can be reduced at the same time as network management efficiency can be increased. It is also an advantage that network management, e.g. operation and maintenance, can be facilitated and improved also for evolving and changing network structures. A particular advantage lies in that it becomes possible to integrate and manage several heterogeneous environments, different network operators and network vendors. A positive aspect is also that in several ways network management can be automated more or less independently of network size and complexity. A particularity of the invention is that the need for human intervention and financial costs associated therewith are reduced. This is achieved through automation of configuration and management upon a grouping structure which is adapted to the requirements driven by the specific configuration and management tasks that need to be achieved. It is also particularly an advantage that it becomes possible to timely, in a managed and controlled way, handle the grouping while eliminating human errors and omissions etc. An important advantage is also that adaption to new requirements or changes of behaviour can be handled without requiring and re-implementation of grouping protocols (which is time-consuming costly and involves a risk of introducing errors into the system).
The invention will in the following be more thoroughly described, in a non-limiting way, and with reference to the accompanying drawings, in which:
According to the present invention policies are used for the purpose of grouping nodes or network elements. Through a policy-based grouping according to the present invention, a plurality of nodes or network elements, for example computers, base stations, routers or resources are enabled to form a self-organizing and adaptive group or groups for the purpose of their configuration and management. The group formation particularly takes place automatically in so far that no or only a minimum of human involvement or interaction is required. According to the inventive concept, nodes or managed network elements particularly are provided with the capability of explicitly controlling their grouping. For this purpose policies are enacted which serve as general guidelines to restrict and define both the behaviour of the group members and the metrics used to create and adapt the group structure according to the current situational context (the network context) and according to the configuration and management tasks that have to be performed.
In a particular embodiment the node comprises a policy engine which comprises monitoring means adapted to collect current and/or change related information concerning status and/or activity or nodes, groups and network and further being adapted to provide said information to a policy decision means. Policy handling means may also be provided which are adapted to indicate and/or activate one or more grouping policies to be used in the node grouping and to provide information about said one or more grouping policies to the policy decision means. The policy decision means are then adapted to interpret said active grouping policies and to make grouping related policy decisions based on said interpretations. The policy engine preferably also comprises policy enforcement means which are adapted to enforce said decisions.
In an advantageous implementation said policy enforcing means are adapted to enforce said decisions using the grouping protocol interacting with the communication layer of the communication protocol. The policy handling means may particularly comprise a policy management tool with a user interface. This enables a network administrator or some management operator to activate, select or indicate policies to be used for grouping purposes. The policy handling means may also, or additionally, comprise or communicate with policy holding means for holding grouping policies. In one embodiment the policy holding means are provided locally on the node. They may also be associated with the node, or associated with some nodes.
In a particular implementation the policy decision means are adapted to control a grouping related decision process, which may relate to one or more of group membership of nodes, group structure, group or group member constraints and metrics. This decision process may for example be used to answer to specific objectives of network configuration management tasks or applications which are provided by a configuration and management application.
The grouping protocol preferably comprises a group management protocol providing messaging capabilities required for group controlling and node group member communication.
Advantageously the policy enforcement means are adapted to control, for example by giving orders to, the grouping protocol, wherein said orders correspond to the decisions made in the decision means.
In a preferred, but optional, implementation, the policy decision means comprises a policy conflict resolution function which is adapted to detect conflicts between policy rules of different active policies and to resolve the conflicts either automatically or by means of invoking (to some extent) manual or operator intervention.
The grouping policies are particularly defined as sets of policy rules. Particularly the node is adapted to comprise or communicate with configuration and management applications, and the policy engine is particularly adapted to create and dynamically adapt the grouping and grouping related features or actions according to current network context and configuration and management tasks provided (or input) by said configuration and management applications.
In a particular implementation each policy rule comprises a set of conditions and a set of actions related to said conditions. A condition may comprise a compound expression which is related to one or more entities such as group members, hosts and grouping protocols. An action relates to a set of actions specifying parameters to be input to one or more grouping protocols.
In the present invention a node may comprise a computer, e.g. a PC (Personal Computer), a PDA (Personal Digital Assistant), a node of a mobile telecommunications network, for example of a radio access network, e.g. a Radio Base Station (RBS), an RNC (Radio Network Controller) or any other node of the radio network or a node of a core network. It may also comprise any node of a data communications network or generally a Network Element (NE) or some kind of a managed node.
Particularly, in an advantageous implementation, the method comprises the steps of; providing the active grouping related policies to be used for grouping purposes to a policy decision means of the policy engine of a node; monitoring in monitoring means provided in the node, current status and/or change related information of said nodes, group and network; providing said current status and/or change related information to the policy decision means; interpreting the grouping policies with respect to current status and/or activity related information; making grouping related decisions based on said interpretations; enforcing the decisions made by means of the grouping protocol interacting with the communication layer to enable interaction and communication with other nodes. Other preferred or advantageous, optional, features may be implemented in any desired manner, also with respect to the inventive method.
In a particular implementation the node comprises a policy conflict resolution function, advantageously implemented in the policy manager 3A, particularly PDP 31, which is adapted to detect conflicts between e.g. contradicting or overlapping policies, for example to prevent the system from any unpredictable or faulty grouping behaviour.
A policy particularly guides the grouping behaviour and regulates how the group is structured through high-level declarative directives corresponding to rules. Each policy rule particularly comprises a set of conditions and a set of related actions. Policy rules are in the form of if “condition” then “action”. The “condition” expression may be a compound expression relating to entities such as group members, hosts or grouping protocols. The “action” expression may relate to a set of actions that specify for example parameters to be input or provided to one or more grouping protocols.
According to the actions given by the policy engine, particularly the policy managing means 3, 3A or more generally the policy engine here comprising this manager, (PDP 31, PEP 32), policy handler 2 or policy handling means 2A (with policy management tool 21A and policy repository 23A) and monitoring means 1A, the grouping protocol can adapt the grouping structure dynamically, substantially without requiring any human intervention at all. Particularly the monitoring means (e.g. 1;1A;1B) do not have to be included in the policy engine, but may consist of separate means.
If a conflict resolution procedure is implemented, the node 10; 10A particularly comprises a conflict detection function which is adapted to verify that policies do not contradict each other. A policy conflict resolution procedure may be automatically invoked when a conflict has been detected, or by means of some kind of intervention when a conflict has been detected and indicated. It may either be done automatically or operator controlled. In some embodiments the intervention of a network administrator may be mandatory to determine which policy is to be used in case two or more policies contradict each other or overlap etc.
Alternatively a conflict is regulated automatically by means of a predetermined scheme where different policies are given different priorities and then only if two (or more) policies have the same priority human intervention is required. It is of course also possible to provide for automatic selection of a policy to be active according to some predefined criteria, (e.g. depending on which policy was activated first, which policy concerns the largest number of entities or arbitrarily, randomly etc.), even when the priority is the same.
It should be clear that of course embodiments are also covered wherein the applications reside within the node. In a similar manner, a policy repository or similar may be outside the node or inside the node, but not related to, or related to, the applications residing in the node. In this case, hence, the policy engine includes the policy managing means 3B, the monitoring means 1B and the policy handling means 2B, whereas in
Below some examples of policies for group formation are given. It should be clear that the examples are given purely for exemplifying, non-limiting, purposes and they should in no way be interpreted in a restrictive manner, i.e. many other policies concerned with group formation or features thereof can be used, either separately or in combination.
Hence, as a first example there may be a policy based on the role and logical hierarchy of the network elements which policy is defined to group all, here, RBSs together and give the leadership of the group to an RNC and further to group neighbouring RNCs. An example of a corresponding rule would be: if the concerned node is an RBS, then notify the nearest RNC and connect to it.
Another policy may be based on geography for grouping network elements according to geographical proximity or similar. Still further a grouping policy may be based on application level so as to group nodes or network elements according to shared interests, or shared information, or one or more common functionalities.
Still another example is a policy relating to group control, for example defining the maximum number of RBSs that can be lead or controlled by an RNC before splitting the group and sharing the leadership with another RNC.
An example of using these policies could be a grouping structure for a configuration and management task corresponding to tracking and managing the events triggered by some nodes or network elements. For this purpose an efficient grouping structure consists in dynamically organizing the nodes or network elements in a tree-like manner in order to allow the nodes of the tree to perform event aggregation and correlation, and therefore, reduce the traffic generated by the event dissemination. Such a grouping can be achieved as follows: Supposing that a management application subscribes to a particular kind of event, the requirement is triggered by the monitoring manager and redirected to the policy decision point. The policy decision point takes into account the role of the local host, for example an RNC or an RBS, and assigns the local node a position in a given level of the tree. The considered number of possible children that are selected depends on the location of the network element or node. After receiving the decision made by the PDP, the PEP enforces it by giving the information in a comprehensive way to the protocol that implements it, i.e. structure the members in a tree-like manner. It should be clear that this merely is an example among many others intended to show how the inventive concept can be implemented.
According to defined or activated policies nodes may, according to the inventive concept, permanently or temporarily organize themselves into groups based on for example the metric or metrics driven by the configuration and management tasks that are to be performed. Therefore, despite a large network size and a large network complexity, many aspects of network management can be automated through a policy based system as discussed above providing an explicit organization of the network used as a basis for performing management tasks.
It is also supposed that more or less in parallel with the provisioning and activation of grouping policies, changes in nodes, in the group and in the network, as well as group characteristics are monitored by monitoring means of the policy engine (alternatively it keeps or collects updated information relating to all nodes implementing the concept), 102. Also the monitoring can take place in different ways, either by continuously monitoring state etc., or by performing the monitoring at discrete steps, or by providing for the ability to detect changes in general or changes of a number of predefined types.
In the policy engine of a node, the activated policies in view of the change related information or current information are used to automatically and dynamically control grouping according to a relevant application, 103.
In one implementation a policy administrator is responsible for keeping the policies that are used by the group members up-to-date. The policy administrator may then use the policy management tool, cf.
Upon reception of such an update, the policy management tool deployed on the receiver updates the policies included in its local policy repository, i.e. it is established if a policy update is received, 400. If yes, the received policy is extracted, 401, and the local policy repository is updated, 402.
It should be clear that the invention is not limited to the specific, illustrated embodiments but can be varied in a number of ways within the scope of the appended claims.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP2006/068448 | 11/14/2006 | WO | 00 | 3/8/2011 |
