Multiplex communications – Network configuration determination – Using a particular learning algorithm or technique
Reexamination Certificate
1999-10-20
2004-08-31
Lee, Andy (Department: 2731)
Multiplex communications
Network configuration determination
Using a particular learning algorithm or technique
C370S390000, C370S400000, C370S408000
Reexamination Certificate
active
06785245
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
This is the first application filed for the present invention.
MICROFICHE APPENDIX
Not applicable.
TECHNICAL FIELD
The present invention relates to multicast services, and in particular to a method and apparatus for controlling fanout at an egress node of a dynamic multicast tree in a differentiated services (DS) network.
BACKGROUND OF THE INVENTION
The recent evolution of communication networks (e.g. the Internet) has increased demand for delivery of dynamic interactive or selective-use differentiated services.
Multicast services include, for example, news or entertainment content, video conferencing for long distance education, or video conferencing for personal or business differentiated services. In this context, a source may be a server at the network edge serving as a source of the content; or an edge node serving as a gateway between affiliated sub-networks, through which the content (which originates elsewhere) is supplied to the sub-network. Similarly, a sink node may be an end-user'communication device (e.g. a personal computer (PC) connected to the network via a modem); or an edge node which serves as a gateway between affiliated sub-networks, and through which the content (having been transported across the sub-network from the source) is supplied to an adjacent sub-network.
In order to enable multicast transport of content, a network provider (NP) must provision the network with both physical and service resources having sufficient band-width capacity. Physical resources are provisioned by installation of the physical plant used to construct the network fabric. Since changes in the physical resources necessarily involves the installation of new hardware, such changes are expensive, and consequently infrequent. Network providers typically utilize various known methods for predicting future bandwidth capacity requirements, and attempt to deploy physical plant having sufficient bandwidth capacity to satisfy anticipated growth in demand for forecast service offerings. However, at least during any specific multicast session, the available physical resources are static, and these may be allocated (typically by allocation of bandwidth and buffers) to multiple differentiated services sessions (or data transport paths) up to a practical limit of the band-width capacity of the physical resource.
The allocation of resources for any communication session is normally conducted during setup of a data transport connection across the network. Various methods are known for service resource allocation, such as, for example, Resource Reservation Protocol (RSVP) and Constrained Routing-Label Distributed Protocol (CR-LDP). In each case, an end-to-end path across the network between source and receiver nodes is first negotiated, and then transport resources along the path are allocated in accordance with a service level specification (SLS) of the communication session. In a situation where the allocation of resources to satisfy the SLS exhausts (or exceeds) the practical capacity of any element (e.g. a node or link) in the path (given pre-existing resource allocations for other differentiated services sessions already utilizing that element), then the path must be abandoned and a new path setup over less congested network elements.
Prior service resource allocation methods are well suited to unicast, and point-to-point (i.e. 1-to-1) connections across the network. In such cases, it is a simple matter to set up a path and allocate resources during setup of the connection. These methods can also be used effectively for multicast groups in which ingress and egress nodes are predetermined and static (i.e. 1-to-Many, in which all egress nodes are predetermined and static) because resource requirements can be accurately predicted prior to setting up the multicast tree. However, in cases where a multicast group is dynamic (i.e. in which any node of the network may serve as an egress node, and/or new egress nodes may join or leave the group during a multicast session), service resources of the network must be provisioned to provide for “1-to-Any” distribution, and it is very difficult to predict provisioning requirements a priori.
In order to overcome this problem Applicant invented a method and system for provisioning network resources for dynamic multicast groups which is described in a Patent Application entitled METHOD AND SYSTEM FOR PROVISIONING NETWORK RESOURCES FOR DYNAMIC MULTICAST GROUPS, issued as U.S. Pat. No. 6,556,544 on Apr. 29, 2003, and is incorporated herein by reference.
Although Applicant'copending patent application addresses the problem of dynamically provisioning a differentiated services network for a dynamic multicast tree, there exists a further multicast provisioning problem that has to date remained unaddressed. That problem is the control of the number of egress points at an ingress node of a dynamic multicast tree. Such control is commonly referred to as the control of “fanout” at the ingress node.
Since an ingress node in a multicast tree has a predetermined bandwidth capacity allocated to any given multicast session, egress points cannot be grafted to the ingress node without regulation. Otherwise, service levels will suffer and the multicast session is likely to intrude on the resources of other services. Egress point grafting to an ingress node of a dynamic multicast tree is not easily controlled, however. The problem is complicated by the dynamic nature of the multicast tree and the unpredictability of how many branches will be required and where the branches are likely to grow.
Consequently, there exists a need for a method and apparatus to permit the control of fanout at an ingress node of a multicast tree. The control is required to enable network service providers to ensure a committed service level in a differentiated services network, while permitting the multicast tree to dynamically respond to access demand within a framework of a service level agreement (SLA) between the service provider and one or more customers that contract for the multicast service.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a method and apparatus for controlling fanout at an ingress node of a dynamic multicast tree in a differentiated services network.
A first aspect of the invention therefore provides a method of controlling fanout at an ingress node of a dynamic multicast tree for a multicast session in a differentiated services network. The method is commenced when a graft request message is received from an egress node that has not been joined to the multicast tree. On receipt of the graft request message, the provider egress router sends a graft-request message to the ingress node. A determination is then made as to whether a number of egress points specified by a service level specification (SLS) associated with the multicast session permits the egress node to be grafted. If the graft is permitted, a graft permission response message is sent to the egress node. A graft is allowed if the current number of egress points at the ingress is less than a maximum number stipulated by the SLS.
If the egress mode is not permitted to be grafted, a graft-redirect message is preferably sent from the egress node to a downstream requesting node to permit the downstream requesting node to attempt to graft to an existing branch of the multicast tree instead The graft-redirect message contains addresses of at least one grafted egress node that is already a part of the multicast tree, each grafted egress node providing an alternate site at which the downstream requesting node can attempt to graft to the multicast tree. The downstream requesting node forwards a graft request message to each of the grafted egress nodes in an order in which the addresses are presented in the graft-redirect message, and the downstream requesting node grafts to a first of the grafted egress nodes that accepts the graft. The downstream requesting node sends only one graft request message at a time and waits for a response from the grafted
Lee Cheng Y.
Seddigh Nabil
(Ogilvy Renault)
Lee Andy
Nortel Networks Limited
Wood Max R.
LandOfFree
Method and apparatus for controlling fanout of a multicast tree does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for controlling fanout of a multicast tree, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for controlling fanout of a multicast tree will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3362648