Routing methods, systems, and computer program products

What is this patent about?

’010 is related to the field of network routing and packet forwarding, specifically addressing the challenges of latency and address space exhaustion in large-scale networks. Traditional routing relies on global address lookups and complex routing tables that must be synchronized across all nodes. The invention seeks to streamline this process by shifting from traditional destination-based addressing to a more flexible, path-oriented approach that utilizes segment-based identifiers to direct traffic through a network.

The underlying idea behind ’010 is to replace or augment traditional global IP addresses with a sequence of local instructions, or segment identifiers, that explicitly define the path a message should take. Instead of each router independently deciding the next hop based on a global destination, the source or a controller pre-determines the route and encodes it as a stack of identifiers. This allows the network to function more like a set of directed instructions, where each node simply follows the directive at the top of the stack to move the packet forward.

The claims of ’010 focus on a method and system for identifying and assembling a plurality of segment identifiers into a segment identifier stack that encodes a specific network path. The process involves accessing a data structure that maps these identifiers to specific nodes, where the identifiers themselves are distributed or updated via network advertisements. The system then inserts this stack into a message header and forwards the message by matching the top segment identifier against a local forwarding table that has been populated through these advertisements.

In practice, the invention works by having a node or controller look up the necessary hops to reach a destination and pushing those identifiers onto a stack within the packet header. As the packet traverses the network, each node examines the top-most identifier to determine the correct outgoing interface. This mechanism effectively offloads the path-selection logic from the core routers to the edge or a centralized controller, allowing for highly granular control over the specific sequence of nodes a message visits.

This approach differs from prior solutions by moving away from the rigid distinction between names, addresses, and routes. Unlike standard IP routing, which requires every router to maintain a massive global routing table, this system uses path-based protocol addresses that are often scope-specific or node-specific. By using a stack of identifiers that can be popped or modified at each hop, the invention enables source-directed routing that is more efficient, scalable, and capable of supporting complex traffic engineering without the overhead of traditional global address management.