摘要:
A technique protects traffic (IP) against the failure of a border router between two domains in a computer network using Fast Reroute and backup tunnels. The border router (i.e., the “protected border router”) announces/advertises a list of all its adjacent next-hop routers (i.e., its “neighbors”) residing in first and second domains interconnected by the protected border router. A neighbor in the first domain that is immediately upstream to the protected border router and that is configured to protect the border router (i.e., the “protecting router”) learns address prefixes (i.e., “protected prefixes”) reachable from the next-hop router in the second domain (i.e., “next-next-hops,” NNHOPs to the protected prefixes from the protecting router). The protecting router calculates a backup tunnel to each NNHOP that excludes the protected border router, and associates each backup tunnel with protected prefixes accordingly. When the protected border router fails, Fast Reroute is triggered, and the protected prefixes are rerouted by the protecting router onto an appropriate backup tunnel to a corresponding NNHOP.
摘要:
A technique efficiently determines acceptable link-based loop free alternates (LFAS) in a computer network. According to the novel technique, a protecting network device configured to protect a link (“protected link”) distinguishes other network devices (e.g., of the same domain as the protecting network device) as either network edge devices (i.e., an end point for external network traffic) or network core devices (i.e., not an end point for external network traffic). The protecting network device may then determine whether a neighboring network device loops toward a network edge device. If not, the protecting network device may determine that the neighboring network device is an acceptable LFA (e.g., for external network traffic). Notably, traffic directed to core devices may still loop, however, this traffic is generally internal (e.g., signaling) traffic, and may not be subject to the same operational constraints (e.g., protection) as external traffic.
摘要:
A technique optimizes routing of application data streams on an Internet Protocol (IP) backbone in a computer network. According to the novel technique, a client router learns of server states (e.g., number of pending requests, etc.) of a plurality of application servers and also determines metrics of intermediate links between the application servers and the client router (intermediate link metrics), e.g., particularly link metrics in a direction from the application servers to the client router. Upon receiving an application request from an application client (“client request”), the client router determines to which of the application servers the client request is to be sent based on the server states and intermediate link metrics, and sends the client request accordingly.
摘要:
A system and method for advertising out-of-resources (OOR) conditions for entities, such as nodes, line cards and data links, in a manner that does not involve using a maximum cost to indicate the entity is “out-of-resources.” According to the technique, an OOR condition for an entity is advertised in one or more type-length-value (TLV) objects contained in an advertisement message. The advertisement message is flooded to nodes on a data network to inform them of the entity's OOR condition. Head-end nodes that process the advertisement message may use information contained in the TLV object to determine a path for a new label switched path (LSP) that does not include the entity associated with the OOR condition.
摘要:
An apparatus and method as described for constructing a repair path for use in the event of failure of an inter-routing domain connection between respective components in first and second routing domains of a data communications network. The apparatus is arranged to assign a propagatable repair address for use in the event of failure of the inter-routing domain connection and to propagate the repair address via data communications network components other than the inter-routing domain connection.
摘要:
A technique calculates a shortest path for a traffic engineering (TE) label switched path (LSP) from a head-end node in a local domain to a tail-end node of a remote domain in a computer network. The novel path calculation technique determines a set of different remote domains through which the TE-LSP may traverse to reach the tail-end node (e.g., along “domain routes”). Once the set of possible routes is determined, the head-end node sends a path computation request to one or more path computation elements (PCEs) of its local domain requesting a computed path for each domain route. Upon receiving path responses for each possible domain route, the head-end node selects the optimal (shortest) path, and establishes the TE-LSP accordingly.
摘要:
In one embodiment, one or more path computation requests from path computation clients (PCCs) in a first network domain are received at a first border router (BR) arranged at the border of the first network domain and a second network domain. The first BR learns of a path communication element (PCE) in the second network domain. The PCE in the second network domain is informed of path computation information for the first network domain. One or more tunnels are established between the first BR and the PCE in the second network domain. One or more path computation requests from PCCs in the first network domain are passed from the first BR, through the one or more tunnels, to the PCE in the second network domain, to be serviced by the PCE in the second network domain using the path computation information for the first network domain.
摘要:
In one embodiment, a first path computation element (PCE) operates between first and second network domains, and is adapted to service requests from path computation clients (PCCs) in at least the first domain. In response to a backup event (e.g., failure of a second PCE), a backup PCE in the second domain may be informed of path computation information for the first domain used by the first PCE, and tunnels may be bi-directionally established between the first PCE and the backup PCE. Once the tunnels are established, the backup PCE may be advertised into the first domain, and the backup PCE may operate to load balance service requests for the first domain through the bi-directionally established tunnels.
摘要:
A technique protects against the failure of a border router between two domains in a computer network using Fast Reroute and backup tunnels. According to the technique, the protected border router advertises a list of all its adjacent next-hop routers (i.e., its “neighbors”). A neighbor in the first domain that is immediately upstream to the protected border router and that is configured to protect the border router (i.e., the “protecting router”) selects a neighbor in a second domain (i.e., a “next-next-hop,” NNHOP) to act as a “merge point” of all the NNHOPs of that domain. The protecting router calculates a backup tunnel to the merge point that excludes the protected border router and associates the backup tunnel with all “protected prefixes.” The merge point then “stitches” additional backup tunnels onto the backup tunnel to provide a stitched tunnel to each remaining NNHOP. When the protected border router fails, Fast Reroute is triggered, and all protected prefix traffic is rerouted onto the backup tunnel to the merge point, which either forwards the traffic to its reachable prefixes or to a corresponding stitched tunnel.
摘要:
In one embodiment, a node receives traffic sent from one or more sources toward one or more destinations (e.g., Multipoint-to-Point, MP2P traffic). The node may detect a burst of received traffic based on one or more characteristics of the burst traffic, and, in response, may dynamically apply traffic shaping to the burst traffic. The traffic shaping is adapted to forward burst traffic received below a configurable threshold at a configurable pace and to drop burst traffic received above the configurable threshold. In addition, the node may also store the burst traffic dropped by traffic shaping, and forwards the stored burst traffic toward its destination after a configurable delay.