Amazon Technologies, Inc.

Mediatek Corporation

Kari Ann O'Brien

Bijendra Singh

Thomas A Volpe

Kiran Kalkunte Seshadri

Nafea Bshara

Mark Bradley Davis

George McNeil Lattimore

Thomas Bradley Scholl

Leonard Thomas Tracy

Paul William Hollis

Sravya Kusam

Anand Narayan Acharya

Matthew B Rutledge

Joern Soersensen

Disclosed are devices, systems, and methods for the use of memory including a data table configured to store a plurality of elements, wherein the plurality of elements are arranged into a plurality of buckets and each of the plurality of buckets comprising a plurality of entries. A first power domain can be associated with an entry of each bucket or with a first bucket. A second power domain can be associated with a second entry of each bucket or a second bucket. Processing logic can be configured to search for a particular value stored in an element of the plurality of elements by selecting buckets of the plurality of buckets and selecting at least one entry of each of the buckets. A programmable register can be used to select a powered state of the second power domain based on a configuration of the programmable register.

Power reduction methods for variable sized tables

A packet processor or packet processing pipeline may implement programmatic selection of load balancing output for forwarding paths. Programmatic selection for load balancing among forwarding paths may be enabled or disabled at one or more stages. If programmatic selection is enabled when network packets are received, a programmatically identified path may be provided as the selected path of a load balancing scheme at a stage. Forwarding of the packet may then be performed according to the selected path.

Programmatic selection of load balancing output amongst forwarding paths

Power management of a routing table, which comprises a plurality of hash tables, is provided by supporting various power domain configurations. Each power domain configuration can be associated with a different number of power domains than other power domain configurations. Each power domain can add vertical capacity to the entire routing table by adding a fixed number of buckets to each of the hash tables. Efficient power management can be achieved by switching between a lower power domain configuration and a higher power domain configuration during runtime based on the dynamic load conditions.

Power management of routing tables using vertical scaling

Packet processors or other devices with packet processing pipelines may implement pipelined evaluations of algorithmic forwarding route lookups. As network packets are received, a destination address for the network packets may be divided into different possible prefix lengths and corresponding entries in a routing table for the different possible prefix lengths may be determined according to a hash scheme for the routing table. The entry values may be read from the routing table and evaluated at subsequent stages to identify the entry with a longest prefix match with respect to the destination address for the network packet. The routing table may include entries for different types of network packets and may be configured to include virtual routing and forwarding for network packets.

Pipelined evaluations for algorithmic forwarding route lookup

Entropy in routing tables may be increase to perform packet forwarding. Hash tables that store forwarding routes may be divided into multiple hash table segments. Forwarding routes may be stored across the hash table segments in different route segments. When looking up route segments to identify a forwarding route for a destination address of a network packet, digest values generated for previous route segments may be used to determine hash key values to identify hash table entries in a different hash table segment. The forwarding route may be identified according to the hash table entries located based on the digest values for previous route segments.

Increasing entropy across routing table segments

Destination resolution stages that determine a forwarding decision for a network packet may be selected. Different resolution stages in a packet processing pipeline may be individually identified. Upon accessing an entry in a lookup table in a stage of the packet processing pipeline, a pointer type of a pointer in the entry may identify a next destination resolution stage for determining the forwarding decision of a network packet. Different types of network packets may have forwarding decisions determined using different destination resolution stages, one or more shared resolution stages, or the same resolution stages.

Determining destination resolution stages for forwarding decisions

A cache-and-overflow memory handles both cache and overflow data in a single hardware structure so as to increase speed and reduce supporting hardware structure needed to implement an effective memory system. A single hash value can be used to access either a cache data value or an overflow data value stored in the combined cache-and-overflow memory. If there are a small number of overflow entries, the combined cache-and-overflow memory provides more availability for cache entries. However, overflow entries are provided priority over cache entries. As a result, the combined cache-and-overflow memory dynamically reallocates its space to efficiently store as much as cache as possible until space is needed for overflow data. At that time, the cache data is evicted in a priority order to make space for the overflow data.

Combined cache-overflow memory structure

Disclosed herein is an apparatus for processing an Internet Protocol (IP) header and label switching (LS) headers of a packet in a pipeline. The apparatus includes an LS header processing circuit configured to select a first operation for the packet using an LS header from the packet, and an IP header processing circuit configured to perform an IP lookup to select a second operation for the packet. The apparatus further includes a tunnel initiation circuit configured to initiate an LS tunnel or IP tunnel. The LS header processing circuit, the IP header processing circuit, and the tunnel initiation circuit are operable to operate sequentially on a same packet and concurrently on different packets in a pipeline. Each of these circuits is operable to be bypassed based on an outermost header in the packet, or the selected one of the first operation or the second operation.

Non-recirculating label switching packet processing

A timer scheduler is used to track timeout values for network connections. A single hardware timer generates timeout values that can be tracked per connection in a linked list that is processed at set time intervals. All tracked connections can have a future timeout scheduled. Future timeout values can be stored in both a linked list and a connection state table that cross-reference each other. The linked list is traversed at predetermined intervals to determine which entries have timed out. For each entry that timed out, a second check is made against a timeout value in the connection state table. If timeout value within the connection state table indicates that a timeout occurred, then the network connection is terminated.

Time-out tracking for high-throughput packet transmission

A system can provide a unified quality-of-service for multi-path label switching (MPLS) traffic. A single network device can be configured as an ingress label switch router, an egress label switch router or a label switching router for different label switching paths. The network device can support uniform deployment model, pipe deployment model or short-pipe deployment model in any configuration. A configurable scheduler priority map can be selected from multiple scheduler priority maps to generate a scheduler priority class which can be used to assign priority and resources for the packet.

Unified quality of service (QoS) for label switching traffic

Power management of a routing table is provided by supporting various power domain configurations. Each power domain configuration can be associated with a different number of power domains than other power domain configurations. Efficient power management can be achieved by switching between a lower power domain configuration and a higher power domain configuration during run-time based on the dynamic load conditions.

Power management of routing tables using horizontal scaling

A method for accessing a memory location may include receiving packet data associated with a network connection. A first hash value and a second hash value are generated using the packet data. The first hash value identifies a hash bucket with a plurality of connection state entries. An index entry of the hash bucket is accessed using the first hash value. The index entry includes a plurality of hash keys for the corresponding plurality of connection state entries. The second hash value is matched with one of the plurality of hash keys in the index entry to obtain an offset. One of the plurality of connection state entries for the network connection is identified based on the first hash value and the offset. The identified connection state entry is accessed.

Accessing a memory location using a two-stage hash scheme

A fast reroute (FRR) implementation utilizes an optimized FRR table to implement a fast switchover to a backup link upon detecting a primary link failure on a label switched path (LSP). The switchover can be performed in a negligible amount of time (close to zero milliseconds) based on a backup indicator flag which can be dynamically updated. The FRR table stores pre-determined backup egress interfaces and FRR labels corresponding to primary egress interfaces to minimize the switchover time.

Robust fast re-routing for label switching packets

One pre-configured code point from of a plurality of code points for a multiprotocol label switching (MPLS) traffic class represents a congestion marked code point associated with traffic congestion and the remaining code points represent not-congestion marked. Congestion associated with a label switching path can be determined based on a congestion notification from a router queue and on a pre-determined congestion indicator from a packet header.

Growing community of inventors

Austin, TX, United States of America

Kari Ann O'Brien