Automatic router configuration

[0001] Invention relates to digital networking apparatus and methodology, and more particularly to automatic router configuration approach for improving network switching performance and flexibility.

[0002] In conventional network systems, substantial and complex effort is required typically to configure network topology for switching and routing equipment. More recently, so-called multi-layer switches provide interconnection at Layer 2 and Layer 3 protocol to improve network integration and performance. However, because such multi-layer switches effectively functioning as hardware-based routers, thereby employing conventional routing protocols such as RIP (Routing Information Protocol) or OSPF (Open Shortest Path First) to calculate paths through the network, network managers still must reconfigure ports and reassign subnetwork numbers. Accordingly, there is need for simplified network router configuration.

[0003] Invention resides in simplified and automated approach to network configuration using improved multi-layer switching apparatus and methodology. In particular, packet flows or signals are learned or “snooped” while passing through improved multi-layer switch, and representations of fast routes are loaded into forwarding table including Layer-2 and/or Layer-3 addressing information in digital storage, such as content-addressable memory (CAM), on flow-by-flow or route-by-route basis. Preferred approach improves performance by avoiding pre-determination of network topology. Improved multi-layer switch is preferably installed flexibly (i.e., plug-and-play), essentially without configuring network or improved multi-layer switch. Such multi-layer switch offloads significant portion of traffic load from routers, thereby freeing routers for other use.

[0005] In FIG. 1A, block diagram of conventional (PRIOR ART) network router 4, such as commercial network routing equipment available from Cisco, couples wide-area network (WAN) interface 2 to one or more 10/100 Ethernet subnet local-area network (LAN) 6. Ordinarily, packets are routed by router 4 to and from various source and destination network nodes coupled thereto.

[0006] In accordance with preferred embodiment of present invention claimed herein, conventional network router system is modified, as shown in FIG. 1B, to include improved multi-layer switching circuit and associated software functionality 8 coupled 10 thereto. As coupled, improved multi-layer switch 8 serves effectively to “front-end” conventional router 4, thereby offloading substantial portion of traffic handled by router 4, extending useful life expectancy of router 4, and improving overall network performance. In particular, simplified and automated network configuration is achieved whereby packet flows are identified while passing through improved multi-layer switch. Additionally, fast packet routes are loaded into storage, such as content-addressable memory(CAM), on flow-by-flow basis. Preferred approach improves performance by avoiding pre-determination of network topology. Also preferably, improved multi-layer switch is installed flexibly (i.e., “plug-and-play” functionality), essentially without configuring network or improved multi-layer switch. Improved multi-layer switch offloads significant portion of router traffic load, thereby freeing router for other use.

[0007] Preferably, Internet Protocol (IP) configuration, according to present invention, employs standard IP protocols, such as Address Resolution Protocol (ARP) and Internet Control Message Protocol (ICMP), thereby providing functional compatibility with standard IP networks. Moreover, by not requiring particular routing protocol, it is contemplated that present configuration approach may be implemented as well in networks using other network standard protocol, such as Routing Information Protocol (RIP), Open Shortest Path First (OSPF), and Interior Gateway Routing Protocol (IGRP, i.e., as specified commercially by Cisco).

[0008] To set-up initially, improved multi-layer switch, as defined herein, is interconnected between WAN interface 2/backbone router 4 and LAN Ethernet subnets 6. Preferably, improved multi-layer switch functionality 8 includes packet processing software code and/or hardware logic, including general or custom system or network processor or controller and associated code or instructions for enabling Layer-2 and Layer-3 switching therebetween, for example, as accomplished substantially using “IMPROVED MULTI-LAYER SWITCHING APPARATUS” specified in co-pending U.S. patent application Ser. No. 09/118,458, filed Jul. 17, 1998, which is hereby incorporated by reference.

[0009] Generally, when operating according to present network configuration technique, improved multi-layer switch 8 automatically determines or learns adaptively and dynamically topology of network coupled thereto (i.e., Layer-2 Media Access Control (MAC) and Layer-3 Internet Protocol (IP) addresses). To determine presence and location of router 4, particularly port addresses, for example, improved multi-layer switch 8 emits request signals, interprets corresponding reply signals, as well as observes routing protocol updates arriving periodically from router 4. Moreover, to determine presence and location of subnetworks in LAN segments 6 attached thereto, improved multi-layer switch 8 monitors and identifies IP source addresses of incoming ARP request and response messages, wherefrom improved multi-layer switch 8 may learn logical connections, or virtual LANs effectively, between each router port and any LAN segments in same subnetwork.

[0010] In particular, by actively generating and monitoring (i.e., effectively snooping) on ARP signals associated with received packets, improved multi-layer switch 8 automatically determines and learns current locations of network endstations, nodes, or ports coupled thereto. Importantly, such result is achieved because network IP devices ordinarily broadcast ARP signals to query and learn MAC addresses of other IP devices. Preferably, improved multi-layer switch 8 imposes certain delay period prior to broadcasting signals to segments within appropriate subnetwork 6, then copies or otherwise provides Layer-2 MAC and Layer-3 IP addresses of sources, along with incoming port numbers, into forwarding table provided in digital storage, such as Content-Addressable Memory (CAM) or other equivalent memory. CAM may be configured for storing separate forwarding tables for Layer-2 and Layer-3 addresses. Moreover, aging-out unused entries in Layer-2 and Layer-3 tables may ensure that forwarding table is current with subsequent changes to network.

[0011] Upon receiving unicast packet from local endstation, improved multi-layer switch 8 initially determines whether packet is being transmitted between subnetworks; if not, packet is switched properly to output port according to packet destination MAC address, in compliance with Layer-2 switching protocol. Non-IP packets and packets coming from router 4 are switched at Layer 2. Further, in case when packet travels between subnetworks 6, improved multi-layer switch 8 looks-up or searches for destination IP address in forwarding table; if no match is found, switch 8 sends packet to router 4. Packet may be bound for wide area or local device which has not yet been learned by improved multilayer switch, in either case, router 4 may forward packet conventionally. When improved multi-layer switch 8 finds match in forwarding table for destination address, switch 8 may perform address substitutions, calculate new IP checksum, manipulation of specified fields (e.g., time-to-live field (TTL)), and send subject packet properly to LAN segment, effectively faster than if performed by router alone. Moreover, in accordance with one aspect of present invention, improved multi-layer switch 8 uses MAC address of appropriate port on backbone router as packet's source address, instead of identifying itself as packet source. In this way, improved multi-layer switch 8 appears transparent to source and destination devices.

[0012] Preferably, present methodology for automatic IP configuration improves network performance by optimizing or maximizing router bypass. Hence, after forwarding table is provided and updated, significant amount of local traffic may be routed or bridged by improved multi-layer switch 8, thereby relieving router 4 of local IP forwarding burden and enabling router 4 to concentrate more on moving wide-area and non-IP traffic. Preferably, network manager need not configure improved multi-layer switch 8 or router 4. Advantageously, improved multi-layer switch 8 approach enables network users and managers practical path to transition from traditional software-based routing to next-generation hardware-based multilayer switching. Furthermore, by using industry networking standards, improved multi-layer switch 8 may be deployed in cooperation with installed networks. Because improved multi-layer switch 8 supports full conventional routing, network manager can begin with automatic configuration scheme according to present invention to achieve relatively fast and easy performance improvement, and as necessary, for example, when system or network application requirements require, activate full routing for more flexible, robust path determination of network configuration.

[0013] During traffic processing, one or more packets are received from router 4 and destined for certain subnet(s). In accordance with present methodology, packet enables improved multi-layer switch 8 effectively to learn that certain host exists on subnet, such that such knowledge is learned and stored in network Layer-2 and/or Layer-3 configuration database or forwarding table in cache device, such as CAM storage. Once such information is stored in CAM, for example, improved multi-layer switch 8 may forward subsequently-received packets destined for same host from other network node or address. In particular, packet traffic received from certain subnet and destined to router 4, according to present methodology, will effectively “hit” or be recognized and directed accordingly by CAM storing or seeded, for example, with Layer-2 MAC address of router 4. This recognition, interception and directed switching scheme allows dynamic interception and control of router bound traffic. Moreover, when Layer-3 IP address entry is provided in CAM storage, then subject packet may be forwarded to specified subnet interfaces according to corresponding forwarding table address; or, optionally, such packet may be attended to by system or network processor CPU for forwarding to subnet 6 or to router 4.

[0014] Preferably, interconnect wiring of improved multi-layer switch 8 and router 4 are physically arranged such that router 4 ports are front-ended conveniently for electrical connectability to odd number port on improved multi-layer switch 8. For example, as shown in FIG. 1C, improved multi-layer switch 8 includes one or more private bridge groups 12 which provide internal Layer 3 interface 13 coupling to neighboring ports 10, 6 for corresponding router and subnet connections. In particular, such port wiring assignments are provided preferably such that corresponding connection to IP subnet may be plugged in physical port having even number, for example, on improved multi-layer switch 8 (Port N+1), relative to corresponding router port assigned to adjacent physical port having odd, consecutively neighboring, port number on switch 8 (Port N).

[0015] In preferred implementation, each internal Layer-3 interface 13 may operate independently from other internal Layer-3 interface 13. As shown in operational diagram of FIG. 2A, significant functional states of internal Layer-3 interface include learn, active, and inactive states, as described in table of FIG. 2B.

[0016] Preferably, at boot time or set-up, improved multi-layer switch 8 causes system processor or equivalent hardware and/or software controller initially to store in forwarding table of cache memory or CAM storage two corresponding physical ports associated with one or more internal Layer-3 interface 13 in private bridge group 12. Thus, when improved multi-layer switch is booted, traffic between router ports and corresponding subnet port are initialized with pre-specified forwarding configuration information, thereby enabling network to continue operation, such as according to certain prior or desired state. In this modular manner, improved multi-layer switch 8 may be installed on “live” and active networks relatively quickly, without effectively interrupting routing or switching traffic flow or system or network application programs.

[0017] In preferred implementation, at boot time, internal Layer-3 interface 13 configures improved multi-layer switch 8 MAC address of odd numbered port into macaddr registers, thereby configuring pair of odd/even ports in private bridge group 12. Also, router 4 odd port is set to forward “unknown” unicast packets to both the subnet port and to network or system processor, and CAM forwarding table for Layer-2 addressing is configured with MAC addresses of such two physical ports (i.e., odd numbered port into macaddr registers) for forwarding packets to network or system processor. Preferably, CAM may not be configured for automatic learning of source MAC addresses on router-and subnet ports, and CAM forwarding table for Layer-3 addressing may be configured with default route to send packets to internal network or system processor port.

[0018] Referring to FIG. 2A, Learn state 16 serves to enable improved multi-layer switch 8 to learn, or attempt to learn, Ethernet MAC address of router 4 attached to odd-numbered Layer-2 port. In this dynamically adaptive or intelligent manner, improved multi-layer switch 8 allows packet traffic to flow unmodified between corresponding pair of ports while switch 8 effectively learns or determines current network topology. Preferably, present methodology for achieving automated network configuration assumes active acquisition or detection from subject packet of router 4 Ethernet MAC address connected to odd numbered port on improved multi-layer switch 8. In particular, Ethernet address enables improved multi-layer switch 8 to configure Layer-2 switch processing logic or code for real-time interception of packets from certain subnet and destined to router 4. When in Learn state 16, CAM entries are provided in forwarding table during IP packet processing. For example, as to packets received on odd-number router port (port P), Layer-2 CAM entry may be established with <destination MAC>, <port P+1>, Layer-3 CAM entry may be established with <destination IP>, <destination MAC>, <port P+1>, system or network processor may store source MAC address, thereby maintaining address list detected on given router port, and such processor may discard subject packet currently being processed. By configuring router/subnet port pair in private bridge group 12, Layer-2 switching logic and/or code may forward unaltered subject packet to corresponding subnet port P+1. Further, by adding Layer-2 CAM entry in forwarding table, subsequent packets on same flow are no longer tracked as destined for “unknown” address, and are processed by switching logic and/or code, thereby reducing load on system or network processor. Moreover, for packets received on even-number subnet port (port P+1), by configuring router/subnet port pair in private bridge group 12, Layer-2 switch logic and/or code forwards unaltered subject packet out corresponding router port P. Preferably, improved multi-layer switch 8 may not be configured to forward packets received on subnet port to system or network processor, whereby improving network performance by avoiding packet processing by such processor. During Learn phase 12, Layer-3 CAM may be provided IP routing information. Without Layer-2 CAM entry intercepting packets from subnet and destined for router, Layer-3 CAM entries are unused; when improved multi-layer switch 8 enters “Active” state 20, such entries are effective.

[0019] Preferably, processing of forwarding table or cache for Layer-2 and/or Layer-3 CAM entries are aged or time-delayed according to present methodology, whereupon during extended period, improved multi-layer switch 8 may exclusively process non-IP traffic from particular MAC address. In this aging scheme, Layer-3 CAM entry may be aged, for example, while Layer-2 CAM entry is refreshed continuously. When Layer-2 entry is active, processor may not obtain subject packet or copy thereof, thereby not re-entering Layer-3 CAM entry. Thus, CAM entries (Layers 2 and 3) may be aged and deleted together. Hence, preferably only Layer-3 aging is performed; once Layer-3 entry is flagged for deletion, corresponding Layer-2 entry is deleted.

[0020] Improved multi-layer switch 8 may convert Layer-3 interface from Learn state 16 into Active 20 or Inactive 22 state upon transition conditions, i.e., router port sampled for new Ethernet addresses for specified delay (e.g., at least 60 seconds); and one or more Source MAC address has been determined 18 on router 4 interface. When improved multi-layer switch 8 determines that internal Layer-3 interface 13 is not connected to single router according to present methodology, such interface 13 enters Inactive state 22, wherein switch 8 continue forwarding traffic between router 4 and subnet ports 6, without attempting Layer-3 routing. Preferably, Layer-2 switching logic and code remains configured with such ports in separate bridge group 12, without attempting Layer-3 operations. Layer-2 source MAC learning effectively offloads system or network processor. Output device, e.g., emitting diode pattern on front panel, may indicate when port pair enters certain state(s). For example, Inactive mode 20 indication facilitates failure condition for equipment repair or removal.

[0021] Internal Layer-3 interface 13 may enter Active state 20 independently of other interfaces 13, for example, when MAC address of router 4 attached to odd-number port is known; or upon determination of single router MAC address attached to router port. When entering Active state 20, Layer-2 CAM is updated with MAC address of router 4 discovered in Learn 16 state of operation, thereby causing traffic originating from subnet port and destined for router 4 to be intercepted and processed by Layer-3 switch logic and code, as described herein. During Active state 20, improved multi-layer switch 8 may attempt to route identified IP flows between different subnet ports, and non-IP traffic are forwarded at Layer-2. IP flows between subnet port and router port may be processed continuously by system or network processor. Preferably, packet processing rules for Active state 20 are substantially same or functionally equivalent to packet processing rules for Learn state 16. Additionally, cache aging process during Active state 20 may be substantially same or functionally equivalent to cache aging process for Learn state 16. Furthermore, switch 8 interface allows user to enable Layer-3 configuration according to present invention, for example, when Layer-3 interface is mapped to only two Layer-2 ports; such Layer-2 ports are physically adjacent (i.e., port numbers 1&2, 5&6, etc.); and lowest numbered Layer-2 port associated with Layer-3 interface is odd-number (i.e., port numbers 1, 3, 5, etc.).

[0022] When operating according to present inventive methodology, IP packets may be communicated between network hosts on different subnets. In FIG. 3A, IP packet is sent between two subnets, whereby Host-A 24 sends packet to Host-B 26. In such representative case, packet-130 is structured as follows:

[0023]

	Source MAC:	Host-A MAC
	Source IP:	Host-A IP
	Destination MAC:	Router-Port-A MAC
	Destination IP:	Host-B IP

When packet-130 is received at Port A+1 46, improved multi-layer switch 8 processes received packet such that destination MAC address of packet effectively “hits” corresponding entry in forwarding table of Layer-2 CAM, since router port-A MAC address was previously provided in Layer-2 CAM, and switch 8 may thereby intercept packets accordingly destined for router 4. When packet represents new packet flow having different addressing scheme, there will not be existing Layer-3 entry in CAM forwarding table, and subject packet may be passed-on to system or network processor for forwarding to network destination location, instead of being handled by hardware in improved multi-layer switch 8. Such processor may forward subject packet out switch port A 42 to router 4 through router port A 38, without further processing. Furthermore, router 4 may process subject packet and send packet-232 through router port B 40 to switch port B 44 with following structure:

[0024]

	Source MAC:	Router-Port-B MAC
	Source IP:	Host-A IP
	Destination MAC:	Host-B MAC
	Destination IP:	Host-B IP

Packet-2 may be forwarded to system or network processor when no Layer-2 CAM entry exists in forwarding table corresponding to destination MAC, and unknown MAC addresses are designated for forwarding to such processor. Simultaneously, subject packet is forwarded by improved multi-layer switch 8 to switch port B+1 as packet-334. Preferably upon receiving packet-232, improved multi-layer switch 8 inserts Layer-2 and Layer-3 CAM entries in forwarding table corresponding to subject packet, according to following rules: Layer-2 CAM entry is created with <Destination MAC>, <port P+1>; and Layer-3 CAM entry is created with <IP>, <Destination MAC>, <port P+1>. Accordingly, after Host-B address information is detected from subject packet analysis, Layer-2 and Layer-3 CAM entries in forwarding tables, such as Destination MAC Address, Output Port, Destination IP Address, Output Port, and Destination MAC address for reformulating packet header (pkt hdr), are updated as shown respectively in FIGS. 4A-B, wherein bold text represents updates, and normal text represents existing entries. After such updates are performed, IP packets from Host-A 24 to Host-B 26 may be routed by improved multi-layer switch 8, and Ethernet packets from router 4 destined to Host-B 26 may be forwarded by improved multi-layer switch 8. Preferably, when improved multi-layer switch routes packet on behalf of router 4, as described herein, Source MAC address of packet is that of router 40 port rather than Source MAC address of switch port 48 previously forwarding such packet(s). When Host-B 26 replies to Host-A 24, improved multi-layer switch 8 processes packets in same or functionally equivalent manner as packets from Host-A 24 to Host-B 24. After first packet is processed, Layer-2 and Layer-3 CAM entries are configured to forward traffic using improved multi-layer switch 8, thereby effectively bypassing processing by router 4.

[0025] In another operating mode according to present inventive methodology, IP packets may be sent by Host-A 24 and destined for Host-W 28 accessible via router 4 WAN interface 36, as shown in FIGS. 3B-C. In such representative case, packet-130 is structured as follows:

[0026]

	Source MAC:	Host-A MAC
	Source IP:	Host-A IP
	Destination MAC:	Router-Port-A MAC
	Destination IP:	Host-W IP

When improved multi-layer switch 8 receives packet-130, destination MAC address hits Layer-2 CAM, because router port-A MAC address is in Layer-2 CAM, and switch 8 may intercept packets destined for router 4. Being a new flow, Layer-3 entry does not exist in CAM forwarding table, and packet is passed to system or network processor for forwarding. Such processor forwards subject packet through switch port-A 42, without further processing, and router 4 processes packet and sends packet through WAN port 36. When response packet is received from Host-W, packet is sent from router port-A 38 to switch port A 42 with following structure:

[0027]

	Source MAC:	Router-Port-A MAC
	Source IP:	Host-W IP
	Destination MAC:	Host-A MAC
	Destination IP:	Host-A IP

Packet is then forwarded to system or network processor due to no existing Layer-2 CAM entry in forwarding table for destination MAC, and unknown MAC addresses are configured for forwarded to such processor. Subject packet is forwarded the packet to switch port A+1 46. When system or network processor receives packet, Layer-2 and Layer-3 CAM entries are provided in forwarding table, according to following rules: Layer-2 CAM entry is created with <Destination MAC>, <port P+1>; and Layer-3 CAM entry is created with <Destination IP>, <Destination MAC>, <port P+1>. Accordingly, after Host-A 24 address information is detected, Layer-2 and Layer-3 CAM entries are updated as shown in FIGS. 4C-D, wherein bold text represents updates, and normal text represents existing entries. After such updates, IP packets from router 4 to Host-A 24 are Layer-2 switched by improved multi-layer switch 8, while system or network processor may continue to be called upon to forward packets from Host-A 24 to Host-W 28.

[0028] Foregoing described embodiments of the invention are provided as illustrations and descriptions. They are not intended to limit the invention to precise form described. In particular, Applicant contemplates that functional implementation of invention described herein may be implemented equivalently in hardware, software, firmware, and/or other available functional components or building blocks. Other variations and embodiments are possible in light of above teachings, such as providing customizable Quality-of-Service (QOS) configuration in router, spanning-tree and bridging configuration from router for Layer-2 switching between subnets, and packet filtering configuration from router for filtering traffic forwarded on behalf of router, and simultaneous interception and processing of packets destined for multiple routers; and it is thus intended that the scope of invention not be limited by this Detailed Description, but rather by Claims following.