Client constituting method

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to communication between network nodes, and in particular to methods and devices that allow network nodes to be automatically configured to be their own default gateway.

【0002】

[Conventional technology]

In the technical field of computer networks, protocol stacks are commonly used to transmit data between network nodes connected by network media such as coaxial cable or twisted pair wiring. Network nodes include devices such as computer workstations, servers, network printers, network scanners, and the like. To harmonize the development and implementation of the protocol stack, the International Organization for Standardization (ie ISO) is called OSI, an acronym for Open System Interconnection, which defines the seven layers of network Puttocol. ) Standard model was promulgated.

【0003】

FIG. 1 is a block diagram of the OSI standard model 10. The model includes hardware layer 12, data link layer 14, network layer 16, transport layer 18, session layer 20, presentation layer 22 and application layer 24. Each layer is responsible for performing specific tasks. Hardware layer 12 is responsible for handling the mechanical and electrical details of the physical transmission of the bitstream. The data link layer 14 is responsible for the generation and decryption of addresses used by hardware protocols and for packet processing, including error detection and correction occurring at the physical layer. For example, the data link layer 14 of an Ethernet network is responsible for generating and decrypting medium access control addresses. Network layer 16 makes connections and routes packets in the communication network, including generating and decoding addresses used by higher-level protocols and maintaining routing information to respond appropriately to changes in load. Responsible for doing, for example, in the TCP / IP protocol, network layer 16 is responsible for generating and decrypting IP addresses.

【0004】

Transport layer 18 is used for end-to-end connections between nodes and message transmission between users in a network, including subdividing messages into packets, maintaining packet order and delivery, flow control, and physical address generation. Responsible for it. Session layer 20 is responsible for implementing the process-to-process protocol. Presentation layer 22 is responsible for resolving formal differences between different sites on the network, including transliteration and multiplexing (echoization). Finally, application layer 24 is responsible for interacting directly with the user. Layer 24 includes, for example, applications such as email, distributed databases, web browsers, and the like.

【0005】

Before the ISO promulgated the OSI standard model, the US Defense Advanced Research Projects Agency (DAPRA) promulgated the ARPNET standard model. The ARPNET reference model includes four layers: the network hardware layer, the network interface layer, the host-to-host layer, and the process / application layer.

【0006】

As the name suggests, the OSI and ARPNET reference models provide guidelines that network communication protocol designers may or may not follow. However, most network protocols define layers that correspond at least loosely to one reference model. In the field of computers, there are many well-known protocols used to transmit data between network nodes. For example, TCP / IP, Apple TalkR (NetBEUI) and IPXX are all well-known protocols used to send data between servers, workstations, printers and other devices connected to a computer network.

【0007】

Even if a network node has a single network interface, it is common for several protocols to work simultaneously within a single network node. For example, a typical computer workstation may use TCP / IP to communicate over the Internet and IPX to communicate with a network server. Similarly, printers may be configured to receive print jobs using either the AppleTalkR protocol or the NetBEUI protocol. Typically, these protocols communicate and interact with lower layer hardware protocols. For example, it is common for two computer systems connected via an Ethernet network to communicate using the TCP / IP protocol. Generally, software routines residing at data link layer 14 or network layer 16 carry data packets between the network adapter and the appropriate protocol stack.

【0008】

Now consider TCP / IP packets transmitted over an Ethernet network. Each packet contains a 48-bit medium access control (called MAC for short) address that identifies another node on the Ethernet network. MAC addresses are commonly known as hardware addresses. The entire Ethernet packet is protected by a periodic redundancy check code (ie CRC code) that is calculated by the source network adapter and packed into the Ethernet packet. The receiving network adapter decrypts the CRC to verify the integrity of the Ethernet packet.

【0009】

If the integrity of the packet is not verified, that is, if an error is detected, the packet is dropped. The IP portion of the TCP / IP protocol, known as a "datagram," is encapsulated within an Ethernet packet. The datagram contains a 32-bit IP address and a 16-bit checksum code that protects the IP header. IP is more commonly known as a network address. If the integrity of the IP header is not verified, the datagram will be discarded. The TCP part of the TCP / IP protocol is encapsulated within the datagram and has a 16-bit checksum code that protects the TCP header and the content of the TCP part of the datagram. If the integrity of the TCP header or TCP part content is not verified, the datagram is discarded and the source then resends the packet if it does not receive an acknowledgment datagram from the intended recipient. It should be noted that this packet contains two addresses, a hardware (Ethernet) address and a network (IP) address. The relationship between these two addresses will be described in detail below.

【0010】

FIG. 2 shows the prior art network 26. Network 26 interconnects network nodes 28, 30, 32, 34, 36, 38, 40, 42 and 44. As mentioned above, a network node is a device such as, for example, a computer workstation, a server, a network printer, a network scanner, and the like. In the following description, it is assumed that the network node has an Ethernet network adapter and carries data using the TCP / IP protocol. Many networks follow a set of standards issued by the IEEE. This set of standards is known in the industry as the IEEE802 family of standards. This specification refers to the IEEE802 family standard.

【0011】

Several network nodes are grouped into one LAN segment via a hub. One because each node in one LAN segment receives a signal when another node attempts to send a packet, and a conflict occurs when two nodes in one LAN segment attempt to transmit a packet at the same time. All nodes in the LAN segment are in a common collision domain. The Ethernet protocol includes a forwarding algorithm that minimizes the possibility of yet another collision when two nodes try to retransmit their respective packets. In Figure 2, network nodes 28, 30 and 32 are joined as LAN segment 48 through hub 46, and similarly network nodes 4, 36 and 38 are joined as LAN segment 50 through hub 50. Network nodes 40, 42 and 44 are coupled as LAN segment 56 via hub 54.

【0012】

Switches and bridges are used to interconnect local or remote LAN segments. Switches and bridges form a single logical network (often referred to as a subnet) to perform operations at data link layer 14 and hardware layer 12 of OSI reference model 10. In Figure 2, switch 58 connects LAN segments 48 and 52. In the Ethernet protocol, packets are addressed by medium access control (ie MAC) addresses. The switch and bridge maintain a list of MAC addresses for network nodes on each LAN segment they connect to. Individual packets are transmitted to the appropriate ports within the switch or bridge that route each packet to a LAN segment that contains a node addressed by the MAC address.

【0013】

A router (routing mechanism) that joins LAN segments so that switches and bridges form subnets, while joining subnets over another network, such as the Internet or a wide area network (WAN). Is used. Routers may also be used to route packets within a common subnet (more on this later). The router maintains a table that associates higher protocol addresses (such as IP addresses) with router ports. In contrast to switches and bridges, routers can see the network as a hierarchical topology. In this case, addresses of large blocks and ranges are sent to other routers for further routing. For this reason, routers are often used to route multiple packets in very large networks such as the Internet.

【0014】

The default gateway is the router to which a node sends a packet when it cannot determine to address the outgoing packet to a node on the same subnet. Packets sent to the default gateway may be processed by some other router before arriving at the destination node.

【0015】

Several protocols have been defined that allow the TCP / IP protocol to operate on lower hardware protocols such as the Ethernet protocol. For example, an Ethernet node configured to use TCP / IP will be assigned a subnet mask and default gateway address to the IP address. The subnet mask identifies IP addresses that are on the same subnet, and the default gateway address identifies routers that handle packets that are not on the same subnet. For packets on the same subnet, Address Resolution Protocol (ARP) is used to find the IP address of the destination node.

【0016】

Suppose node 28 has an IP address of 192.44.133.13 and node 30 has an IP address of 192.44.133.25. Further assume that the subnet mask of node 28 is set to 255.255.255.0. In order to send the packet to node 30, node 28 first performs a bitwise AND operation on the IP address of node 30 and the subnet mask, and compares the result with the bitwise AND operation on the IP address of node 28 and the subnet mask. .. If the two AND operations match, node 30 is on the same subnet as node 28, and the MAC address of node 30 can be determined using ARP. Node 28 then sends out a broadcast Ethernet packet using its own MAC address and node 30's IP address according to ARP. The Ethernet protocol supports unicast and broadcast packets. Broadcast packets are addressed to and received by all nodes on a subnet, while unicast packets are addressed to and received by a particular node. It is a thing.

【0017】

Node 30 responds to this message by sending a unicast packet containing the MAC address of node 30 back to node 28. Node 28 then transmits a TCP / IP packet to node 30 using the MAC address received from node 30. In addition, the node caches this information for future transmissions, minimizing the need to iteratively find the MAC address of a node on the same subnet.

【0018】

Further assume that node 28, configured as described above, attempts to send a packet to node 40, which has an IP address of 168.45.198.2. Node 28 determines that node 40 is not on the same subnet, based on the subnet mask that uses the bitwise AND operation described above. In such cases, node 28 to router 62, which is the default gateway used by node 28 to carry packets addressed to the node that cannot be verified that node 28 is on the same subnet. Relay the packet. The router 62 relays the packet to the router 68, the router 68 to the router 69, and the router 69 to the router 60 in order. Since Router 60 is on the same subnet as Node 40 (which is the destination node), Router 60 uses ARP to discover the MAC address of Node 40 and readdress the packet with the MAC address of Node 40. As a result, the packet is transmitted to the LAN segment 56, and the packet is received by the node 40.

【0019】

Various protocols are defined that allow a node to be configured with an IP address, subnet mask, default gateway address and other parameters. In the following description, the terms "client", "client node" and "node to be configured" mean a network node that wants to get the configuration parameters for itself. Also, the term "server" means a network node that provides configuration parameters. One simple protocol is the Reverse Address Resolution Protocol (abbreviated as RARP), which runs over Ethernet networks and translates Ethernet addresses into IP addresses. When the client broadcasts an Ethernet packet using its MAC address, the server responds by transmitting a unicast packet containing the client's IP address to the client. RARP is primarily used by diskless network nodes.

【0020】

Another standard protocol commonly used in networks consisting of diskless nodes is the bootstrap protocol (BOOTP for short). BOOTP uses the User Datagram Protocol (UDP for short). BOOTP determines the pathname that points to the node's own IP address, subnet mask, default gateway address, BOOTP server address, and boot files loaded from the BOOTP server when the node boots. Allows you to. BOOTP also defines a BOOTP relay agent that allows a node looking for configuration parameters to be serviced by a BOOTP server on another subnet. BOOTP is defined in RFC 951.

【0021】

Also, the Network Information Protocol (NIP for short) is a "polling / defense" in which a network node determines its own IP address and the network node is polled to determine if the IP address is in use. "It makes it possible to identify available IP addresses by using a mechanism. The client protects the IP address in use by responding to the polling message. When a client sends a message seeking an IP address, the NIP server returns a set of IP addresses available on the network. The client chooses one of the IP addresses and uses ARP to do a double validation to make sure that IP address is available. The client then secures its IP address by responding to the NIP server, which records the assigned IP address.

【0022】

Finally, the Dynamic Host Configuration Protocol (called DHCP for short) is an extension of BOOTP described in RFC 1541 and RFC 1533. Being an extension of BOOTP, DHCP provides all the services provided by BOOTP. DHCP is chosen by the network administrator for auto-assignment of IP addresses with finite leases (and thus maturity after a certain period of time), auto-assignment of IP addresses with infinite leases (and therefore never maturity), Can provide static allocation of IP addresses.

【0023】

Table 1 below lists the main messages exchanged according to the DHCP protocol along with a description of the client / server interaction.

【0024】

[table 1]

<img file="JP2000041069A_D0001.tif" />【0025】

The configuration interaction between the client and server begins with the client broadcasting a DHCPDISCOVER message on its local subnet. The BOOTP relay software agent on that subnet delivers the message to a DHCP server that is not on the same subnet. Each server then responds with a DHCPOFFER message containing the available network IP address and other configuration parameters. The server unicasts DHCPOFFER messages to its clients if possible, and uses BOOTP relay agents as needed. Alternatively, the server can also use the broadcast address to broadcast a message to its client's subnet.

【0026】

The client then receives one or more DHCPOFFER messages from one or more servers. The client may choose to wait for multiple responses. Select one server to request configuration parameters based on the configuration parameters offered in the DHCPOFFER message. The client broadcasts a DHCPREQUEST message that always contains a server identifier option indicating which server was selected. This message may also include other options that specify the desired configuration value.

【0027】

All servers receive a DHCPREQUEST broadcast from the client, and any server not selected by the DHCPREQUEST message uses that message as a notification that the client has declined the offer for that server. The server selected by the DHCPREQUEST message commits the connection with the client and responds with a DHCPACK message containing the configuration parameters. If the selected server cannot satisfy the DHCPREQUEST message, the server responds with a DHCPNAK message that denies the client's request.

【0028】

When the client receives the DHCPNAK message, it redoes. When the client receives a DHCPACK message containing the configuration parameters, the client performs a final validation on the parameters to verify the length of time the IP address is valid (lease duration) and the lease identity specified in the DHCPACK message. To do. At this point, the client consists of the client IP address, subnet mask and default gateway address. If the client detects a problem with a parameter in a DHCPACK message, the client sends a DHCPDECLINE message back to the server and repeats the configuration process. The client can waive a lease on an IP address by sending a DHCPRELEASE message to the server. The same is true for message interaction in the BOOTP protocol.

【0029】

[Problems to be Solved by the Invention]

The prior art node uses a subnet mask to determine if the destination node is on the same subnet. The subnet mask mechanism is inflexible because node addresses in powers of 2 must be added to the subnet, which is not always practical or possible. Although it is desirable to configure a node to be its own gateway, prior art configuration protocols such as DHCP do not allow a network node to be configured as its own gateway. Therefore, there is a need to solve this problem.

【0030】

[Means for solving problems]

The present invention provides a configuration agent that allows a network node to be automatically configured with an IP address and a default gateway address so that it can be configured as its own gateway address. .. By configuring the node to be its own default gateway, the node can run a common default gateway routine that handles in-subnet and out-subnet packets. A configuration agent according to the present invention has two networks: one network segment containing the nodes to be configured and another network segment containing a server capable of automatically providing configuration parameters. Placed on top of network equipment (such as switches or bridges) connected to the segment.

【0031】

In the first aspect of the present invention, the constituent agent acts as a snooping agent. A message from the configuration server to the node to be configured is snooped to find the message containing the IP address and default gateway address. Such a message is configured by modifying the IP address offered to the node seeking configuration to copy to the default gateway address and sending the message to its intended destination. It allows the node that wants to be configured to be its own default gateway. In some cases, the message from the node to be configured to the configuration server is modified to ensure that the message from the configuration server to the node seeking to be configured is a broadcast message.

【0032】

In the second aspect of the present invention, the constituent agent acts as a surrogate agent. From the perspective of the node being configured, the surrogate agent appears to be a configuration agent. From the perspective of the configuration server, the surrogate agent appears to be a relay agent if the configuration server and the node it is seeking to configure are on different subnets. If they are on the same subnet, the surrogate agent processes the message to the node requesting to be configured from the configuration server, as in the case of the snooping agent.

【0033】

When the node seeking to be configured sends a message stating that it is requesting to be configured with an IP address and default gateway, the surrogate agent receives the message. The surrogate agent modifies the message to request a unicast or broadcast response from the server. In addition, if the node to be configured and the configuration agent are not on the same subnet, the message is modified so that the configuration server considers the surrogate agent as a relay agent.

【0034】

When the configuration server sends a message to a node that should be configured (perhaps with the surrogate agent as a relay agent), the surrogate agent intercepts the message and copies the IP address provided in the message to the default gateway address. To do. This allows the node that wants to be configured to be configured as its own default gateway. The surrogate agent also replaces the IP address of the actual configuration server with the IP address of the surrogate agent, allowing the node seeking to be configured to consider the surrogate agent as a configuration agent.

【0035】

BEST MODE FOR CARRYING OUT THE INVENTION

The present invention is a configuration agent located on a network device such as a switch. The configuration agent communicates directly with nodes on the same subnet, even if the subnet mask indicates that the nodes are on different subnets, so the network node looking for configuration parameters is its own default. Allows it to be configured as a gateway. Before making a detailed description of the present invention, we consider recent trends in the technical field of computer networks.

【0036】

Traditionally, large TCP / IP networks have often used routers to carry packets between network nodes. However, compared to switches, routers are relatively slow and processor-intensive. In addition, routers are considerably more expensive than switches.

【0037】

As mentioned above, assuming that the subnet mask of the source node indicates that the two TCP / IP nodes connected through the Ethernet network have the destination IP address on the same Ethernet network. You can communicate directly without a router. The source node performs a bitwise AND operation on the subnet mask and its IP address and compares the result with the bitwise AND operation on the subnet mask and its destination IP address. If the results match, the source node translates the IP address to an Ethernet address before sending a TCP / IP packet using Address Resolution Protocol (ie ARP). In the following description, the term "hardware address" means a low-level address used by network hardware such as an Ethernet MAC address, and the term "network address" means a high-level address such as an IP address. Means the address used in the protocol stack.

【0038】

Subnet masking is a relatively unfinished mechanism. IP addresses are added to the network in units of powers of 2. Therefore, if the subnet mask is set to 255.255.255.0, 1024 unique IP addresses will be available on the network and the nodes assigned these addresses will be directly assigned without the need for a router. Can communicate with each other. However, if the network administrator wishes to add more IP addresses, erasing one of the bits in the subnet mask will provide 2048 unique addresses. Not all of these addresses are available in many situations. Therefore, the network administrator should configure TCP / IP to send packets to the router identified by the default gateway address, even if the nodes addressed by the destination packet are on the same subnet. Is forced. In fact, to minimize configuration issues, many network nodes are simply configured to send all TCP / lP packets to the default gateway.

【0039】

One solution to this problem is to configure each node's default gateway address with its own IP address. In this way, when the network node attempts to send a TCP / IP packet, the gateway subroutine on the network node routes the packet. The gateway subroutine uses ARP to find the MAC address to which the packet should be sent. It should be noted that routers and switches can be configured to include ARP surrogate servers. For switches, the ARP surrogate server reduces network traffic by caching the MAC-to-IP address correspondence of all nodes on the subnet. Therefore, when a node attempts to detect the IP address of a remote node on the same subnet and broadcasts an ARP request packet, the packet is simply sent within its LAN segment. No ARP surrogate server is required if the LAN segment is integrated into the subnet through a switch. However, without an ARP surrogate server on the switch, ARP broadcasts are carried to all nodes on the subnet, increasing network traffic. Assuming the switch has an ARP surrogate server, when the gateway subroutine uses ARP to discover the MAC address of a remote node on another LAN segment, the ARP surrogate server provides the MAC address of the remote node.

【0040】

In contrast, if nodes on different subnets communicate using ARP, an ARP surrogate server on the router is required. The ARP surrogate server of the router maintains a correspondence between MAC addresses and IP addresses. However, for out-subnet nodes, the MAC address provided by the router's ARP surrogate server is the router's own MAC address. When a gateway subroutine uses ARP to find the MAC address of a remote node on another subnet, the router's ARP surrogate server responds with the router's MAC address, so packets are sent to the router for further routing. Will be done.

【0041】

ARP is used to discover all IP addresses, so all packets on the same subnet are routed by the switch, and the router only carries packets that are actually away from its Ethernet network. used. Switches are faster than routers, so TCP / IP packets sent within the network are transmitted relatively quickly. Compared to the subnet mask mechanism, gateway subroutines are not limited to adding IP addresses in powers of 2. Also, the range added does not have to be continuous. Therefore, the network administrator has more flexibility in assigning IP addresses to the Ethernet network.

【0042】

One problem faced by network administrators when pursuing this approach is to configure each network node to be its own gateway. Of course, this can be done manually, but it's not a good solution for novice users. In addition, for networks with many nodes, this is not an acceptable solution.

【0043】

To simplify network management, many network administrators have already migrated to automatic host configuration protocols such as BOOTP or DHCP. Unfortunately, these protocols do not support configuring network nodes to be their own gateways. It is unlikely that the network administrator will go back to manual configuration after embarking on the task of establishing an automatic host configuration protocol server and configuring all network nodes to receive configuration parameters from that server.

【0044】

The present invention solves this dilemma. In one embodiment, the switch (or similar network equipment) comprises a DHCP snooping agent. The snooping agent snoops on packets transmitted through the switch and monitors DHCP configuration messages. When detecting a message containing configuration parameters in the direction of a network node, the snooping agent copies the field containing the candidate IP address to the field containing the default gateway. The snooping agent then sends the packet to the intended node. This allows the IP address of the default gateway stored in the network node to be set to the IP address of the node itself.

【0045】

In a second, more sophisticated embodiment, the switch (or similar device) comprises a DHCP surrogate agent. To the network node looking for configuration information, the surrogate agent appears to be a DHCP server. From the perspective of the DHCP server, the surrogate agent appears to be a DHCP / BOOTP relay agent if the client to be configured and the DHCP server are on different subnets. If the client and server are on the same subnet, the surrogate agent processes the server-to-client message in the same way as a snooping agent. DHCP / BOOTP relay agents are known in the industry and are used to relay configuration parameters between networks. The surrogate agent is located between the network node seeking configuration parameters and the DHCP server, so the surrogate agent communicates with the DHCP server by copying the field containing the candidate IP address to the field containing the default gateway. You can change the message that is being sent. Details of the above two embodiments are described below.

【0046】

The present invention is described with reference to the Dynamic Host Configuration Protocol (ie DHCP). However, the present invention is not limited to DHCP. Those skilled in the art will appreciate that the present invention is applicable to other similar protocols such as BOOTP. The DHCP message formats are listed in Table 2 below. This format is taken from RFC1541. The definition of each field is shown in Table 3 below.

【0047】

[Table 2]

<img file="JP2000041069A_D0002.tif" />【0048】

[Table 3]

<img file="JP2000041069A_D0003.tif" />【0049】

FIG. 3 shows a network 74 including a switch 76 with a configuration agent according to the present invention. FIG. 3 is common to both the first embodiment in which the configuration agent is a DHCP snooping agent and the second embodiment in which the configuration agent is a DHCP surrogate agent. Switch 76 connects LAN segment 78 to LAN segment 82 via hub 80 and hub 84. LAN segment 78 includes a network node 86 that is required to be configured, and LAN segment 82 contains a DHCP server 88 that provides configuration parameters. In both configurations of the present invention, the configuration agent on the switch modifies the interaction between node 86 and server 88. Therefore, the switch 76 is preferably arranged between the node 86 and the server 88 so that the communication path between the node 86 and the server 88 passes through the switch 76.

【0050】

FIG. 4 shows an N-port switch 90 according to a first embodiment of the present invention. Each port has an input path to the sending device 100 and an output path that includes output buffers such as output buffers 92, 94, 96 and 98. Further, the sending device 100 includes a DHCP snooping agent 102. Except as described below, the sending device 100 sends out packets in a manner known in the art.

【0051】

FIG. 5 is a flow chart 104 showing how the DHCP snooping agent 102 processes packets. The packet is received at block 106. Block 108 determines if the packet is an HCPDISCOVER, DHCPREQUEST, DHCPOFFER or DHCPACK packet that specifies a network node that should be configured as its own gateway. Note that in one configuration of the invention, switch 90 may be configured to simply "help" a node with a given MAC address by maintaining a table of nodes to be "helped". There is a need. In other words, some nodes should be configured as their own gateway, while others should be configured according to prior art. If the packet is not one of the above types, or if the packet is from a node that should not be "assisted", block 108 branches to "NO" and the packet is transmitted in block 110. Will be done. Control passes to block 106 and waits for the next packet.

【0052】

If the packet is a DHCPDISCOVER or DHCPREQUEST packet, block 108 branches to "YES" and proceeds to block 112, where it is checked to see if the packet's B flag is set. The function of the B flag is defined in RFC1541 as follows.

【0053】

255) may be sent as an IP broadcast using a link layer (Ethernet) broadcast address as the link layer (Ethernet) destination address. Some switches can trap and process unicast and broadcast packets. This type of switch skips blocks 112 and 114, sends a packet at block 110, and clears the B flag. However, other switches can only trap packets that are addressed directly to the switch MAC address. For this type of switch, it is important that all packets involved in the configuration interaction are Ethernet broadcast packets and that the switch CPU can intercept the packets.

【0054】

Initial DHCPDISCOVER and DHCPREQUEST packets are typically Ethernet broadcast packets because the client usually does not know the IP address at the DHCP server. However, if the client can receive a unicast IP datagram before it is configured with an IP address, the client sets the B flag and B as a hint when deciding how the DHCP server sends a response to the node. Flags can be used. When unflagged, DHCPOFFER, DHCPACK and DHCPNAK packets become unicast back to the node, so the agent 102 cannot intercept the packet unless the switch can trap the unicast packet. It is likely that the B flag has already been set, as many nodes will not be able to receive unicast IP datagrams until they are composed of IP addresses. If the B flag is set, it proceeds to block 110, the packet is sent, and control returns to block 106 to wait for the next packet. If unflagged, proceed to block 114, set the B flag, and recalculate the Ethernet checksum. Control moves to block 110, a packet is sent, and control returns to block 106 to wait for the next packet.

【0055】

At block 108, if the packet is a DHCPOFFER or DHCPACK packet, control proceeds from the downward branch "YES" to block 116. Block 116 copies the candidate IP address of the YIADDR field to the default gateway field in the options field, configures the network node to be its own gateway, and then recalculates the packet checksum. Control moves to block 110, a packet is sent, and control returns to block 106 to wait for the next packet.

【0056】

The main advantage of this embodiment of the present invention is that it is easy to implement and requires little overhead. However, this embodiment has a few minor restrictions. For switches that cannot intercept unicast packets, when the node seeking configuration knows the address of the DHCP server, the node uses that address to unicast DHCPDISCOVER and DHCPREQUEST packets. Maybe. Since the initial DHCP packet is not broadcast, the DHCP snooping agent cannot miss such a packet and set the B flag. However, in most implementations of the DHCP protocol, the node seeking to be configured does not know the address of the DHCP server, thus ensuring that the initial DHCP packet is not broadcast.

【0057】

Similar problems can occur when a network node attempts to renew the lease term for its IP address. Since the node knows the address of the DHCP server, it typically unicasts the DHCP REQUEST packet to the server. The DHCP snooping agent resides on a switch that cannot intercept unicast packets, so it sends this message directly to the server without the opportunity to set the B flag. Since the client consists of an IP address and can therefore receive unicast IP datagrams, it is likely that the client will clear the B flag and the DHCP server will respond with a unicast message. The DHCP server is unaware of the modified default gateway address and may attempt to drop the packet or return the default gateway address to an address that it believes is correct.

【0058】

Since the message is unicast to the client, the DHCP snooping agent has no chance to modify the default gateway address. Therefore, if the DHCP server is not modified, this embodiment may have dynamic address allocation issues for switches that cannot intercept unicast packets. On the other hand, if the DHCP server is modified to use the default gateway provided by the client, there is a great advantage that the present invention operates with the DHCP server of the prior art. Of course, switches that can intercept unicast packets are unaffected.

【0059】

Another minor drawback is that setting the B flag will increase network traffic slightly in configurations that would otherwise leave the B flag unflagged. Despite these minor drawbacks, DHCP snooping if the network administrator requires the node to be configured as its own gateway using a prioritized auto-configuration protocol such as DHCP or BOOTP. Agents provide effective and easy-to-configure solutions for many networks.

【0060】

A second embodiment of the present invention does not suffer from the minor drawbacks described above. FIG. 6 shows an N-port switch 118 according to a second embodiment of the present invention. Each port has an input path to the sending device 120 and an output path that includes output buffers such as output buffers 122, 124, 126 and 128. Further, the sending device 120 includes a DHCP surrogate agent 130. The DHCP proxy agent 130 is configured to appear to the node seeking to be configured as the DHCP server 132. In addition, the DHCP / BOOTP relay snooping agent 134 of the DHCP proxy agent 130 is configured to look like a server as a DHCP / BOOTP relay agent if the DHCP server is not on the same subnet. If the client to be configured and the DHCP server are on the same subnet, the DHCP / BOOTP relay snooping agent 134 acts as a snooping agent (similar to the first embodiment) for packets transmitted from the DHCP server to the client. To do. Except as described below, the sending device 120 sends out packets in a manner known in the art.

【0061】

7 to 10 are flow charts 136 showing a method in which the DHCP proxy agent 130 processes packets. Packets are received at block 138. Control passes to block 140, where it determines if the packet is an HCPDISCOVER, DHCPREQUEST, DHCPOFFER, DHCPACK or DHCPNAK packet that specifies a network node that should be configured as its own gateway. As with the DHCP snooping agent 102 described above, the DHCP surrogate agent 130 can be configured to maintain a table of MAC addresses that identify network nodes to be "assisted". If the packet is not one of the above types, or if the node should not be "assisted", block 140 branches to "NO" and the packet is transmitted in block 142. Control then returns to block 138 and waits for the next packet.

【0062】

It should be noted that the DHCP proxy agent 130 receives all broadcast packets transmitted in the configuration dialogue. In addition, if the packet is unicast from a client, the DHCP surrogate agent 130 receives those packets because the client seeking to configure considers the DHCP surrogate agent 130 to be that client's DHCP server. ..

【0063】

If the packet is a DHCPDISCOVER or DHCPREQUEST packet, control proceeds from "YES" to the right of block 140 to block 144. Block 144 determines if the packet is a DHCPREQUEST packet with an IP address in the CIADDR field. If the client node wants to validate the previously assigned configuration parameters, the CIADDR field in the DHCPREQUEST packet should have been filled in by the client node, as shown in Tables 2 and 3 above. .. This process is known in the industry as lease renewal. If the packet is a DHCPREQUEST packet with an IP address in the CIADDR field, control proceeds to block 145.

【0064】

Block 145 determines if the DHCP server and switch are on the same subnet. This can be determined by configuring the DHCP proxy agent 130 with a DHCP server, or the switch can dynamically know the presence of a DHCP server when it receives a DHCP response. If the DHCP server and switch are on the same subnet, control moves to block 146. Block 146 first accesses the server allocation table. The server allocation table maintains a correspondence between the previously configured client and the DHCP server that provided the configuration information. It should be noted that the server allocation table is not always necessary in a network with only one DHCP server. One entry is stored in the table when DHCPACK and DHCPREQUEST packets are processed (more on this later).

【0065】

Block 146 retrieves the address of the DHCP server that made up the client from the server allocation table. Since the server and client are on the same subnet, the address is either a MAC address or an IP address (translated to a MAC address using ARP). Block 146 then readdresses the packet to the MAC address of the retrieved DHCP server. Block 146 can then set the B flag (if not already set) so that the DHCP server responds to the intercepted broadcast message, as in the first embodiment above. Guaranteed. Finally, the packet checksum is recalculated and control passes to block 142.

【0066】

At block 145, control is transferred to block 147 if the DHCP server and switch are not on the same subnet. Block 147, as in block 146, accesses the server allocation table and retrieves the IP address of the DHCP server that made up the client, along with the default gateway address provided by the server. Block 147 then readdresses the packet to the IP address of the DHCP server, adds the IP address of the DHCP surrogate agent 130 to the GIADDR field, clears the B flag (if not already cleared), and checkssum. Is recalculated. By adding the IP address of the DHCP surrogate agent 130 to the GIADDR field and clearing the B flag, the DHCP surrogate agent 130 looks like a DHCP / BOOTP relay agent to the DHCP server, so the unicast packet being replied to is a DHCP surrogate. Returned to agent 130. Control then returns to block 142, where packets are transmitted to the DHCP server through the default gateway of DHCP proxy agent 130. Control moves to block 138 to wait for the next packet.

【0067】

At block 144, if the packet is not a DHCPREQUEST packet with an IP address in the CIADDR field, control proceeds to block 148. Block 148 determines whether the packet is a DHCPREQUEST packet or a DHCPDISCOVER packet. (Here, the DHCPREQUEST packet does not have an address in the CIADDR field, of course, because it is "YES" in block 144 if it has an address in the CIADDR field). If the packet is a DHCPREQUEST packet, the packet contains the parameters requested by the node seeking to be configured and control proceeds to block 150. Block 150 updates the server allocation table with the offered network node-to-DHCP server correspondence. This table is used when a client requests a DHCP REQUEST packet to renew the lease term for an IP address. Control then shifts to block 152. At block 148, if the packet is a DHCPDISCOVER packet, control proceeds to block 152.

【0068】

Block 152 determines if the DHCP server and switch are on the same subnet. This determination is made in the same manner as in step 145 described above. If the switch and DHCP server are not on the same subnet, control proceeds to block 154. At block 154, the IP address of the DHCP surrogate agent 130 is inserted in the GIADDR field, which makes the DHCP surrogate agent 130 look like a DHCP / BOOTP relay agent to the DHCP server. Block 154 then clears the B flag so that unicast transmission is required. This saves network bandwidth. The packet is then addressed with the IP address of the DHCP server. Since the DHCP proxy agent 130 acts as a DHCP / BOOTP relay agent as far as the out-of-subzone DHCP server is concerned, the DHCP server address manual operation that the DHCP proxy agent 130 dynamically discovers the DHCP server by examining the response packet. A list of DHCP servers can also be maintained using prior art such as configuration. Finally, the packet checksum is recalculated and control returns to block 142, where the packet is transmitted through the default gateway of DHCP proxy agent 130. Control then moves to block 138 to wait for the next packet.

【0069】

In block 152, if the DHCP server and switch are on the same subnet, it goes to block 156 and requests broadcast transmission from the DHCP server, so the B flag is set and the checksum if the B flag has already been cleared. Is recalculated. Control then returns to block 142, where the packet is transmitted. Control moves to block 138 to wait for the next packet.

【0070】

On the other hand, in block 140, if the packet is a DHCPOFFER, DHCPACK or DHCPNAK packet on a network that is configured to be "assisted", control goes to block 158 to determine if the packet is a DHCPAC packet. .. If it is a DHCPAC packet, it means that the packet contains configuration information, control proceeds to block 160, and the server allocation table is updated with a network node-to-DHCP server correspondence as in step 150. Control then shifts to block 162. Control is transferred to block 162 even if the packet is not a DHCPAC packet in block 158.

【0071】

At block 162, the IP address of the DHCP surrogate agent 132 is copied into the SIADDR field, which allows the network node seeking to be configured to consider the DHCP surrogate agent 132 as its DHCP server. Block 162 clears the GIADDR field so that the DHCP surrogate agent 132 does not appear to the client node as a DHCP / BOOTP relay agent. The candidate P address is then copied from the YIADDR field to the default gateway field in the options field to configure the node to be its own default gateway. Finally, the packet checksum is recalculated and control returns to block 142, where the packet is sent to the client to which it should be configured. Control then moves to block 138 to wait for the next packet.

【0072】

A second embodiment of the invention provides a comprehensive solution for network administrators seeking to support all aspects of DHCP messages, including lease renewal and unicast transmission, while maintaining a prior art DHCP server. provide. In addition, the second embodiment of the present invention is sufficiently functional when used with respect to a switch capable of intercepting only broadcast packets.

【0073】

By allowing a network node to be automatically configured as its own gateway, the above two embodiments of the present invention maintain automatic configuration and significantly increase traffic through the router. Allows network administrators to extend the network by adding switches and bridges without increasing to. Because switches and bridges are much faster and cheaper than routers, the present invention provides network administrators with the opportunity to minimize the cost of extending the network, maximize the speed of the network, and maintain automatic configuration.

【0074】

In the prior art, network administrators had three choices when expanding a network by adding switches or bridges. First, the administrator was able to change the subnet mask. This choice allows automatic configuration, but often requires adding IP addresses to the subnet with a power of two that is not practical. Second, the network administrator could manually configure each client to be its own default gateway, but this choice allows the administrator to already establish an automatic configuration protocol such as DHCP. If you're spending money on it, you're not an attractive candidate. Finally, network administrators could use routers, but routers are slow and expensive. The present invention provides a fourth option. According to the present invention, the configuration agent allows any IP address increment to be added to the subnet without the need for manual configuration implementation and without resorting to routers.

【0075】

Although the present invention has been described above with reference to a specific embodiment, those skilled in the art will recognize that various modifications can be made to the above embodiment without departing from the idea of the present invention. There will be.

【0076】

The present invention includes, for example, the following embodiments.

【0077】

(1) A method of configuring a client to be its own default gateway, which is part of the configuration interaction between the configuration server and the client and is currently or will be used by the client. A client configuration method including a step of intercepting a message including a default gateway address, a step of modifying the default gateway address in the above message to be the address of the client, and a step of transmitting the above message. ..

【0078】

(2) The client configuration method according to (1) above, wherein the above message is an offer message.

【0079】

(3) The client configuration method according to (1) above, wherein the above message is an acknowledgment message.

【0080】

(4) The client configuration according to (1) above, wherein the step of intercepting a message includes a step of receiving a message from the server, and the step of transmitting a message includes a step of transmitting the message to the client. Method.

【0081】

(5) Modifying the default gateway address in the above message The above step sets the candidate network address that is currently used or will be used as the network address of the client to the client network address of the message. The client configuration method according to (1) above, comprising copying from the field and storing the candidate address in the default gateway address field of the message.

【0082】

(6) The client configuration method according to (5) above, which further includes a step of recalculating the checksum associated with the above message.

【0083】

(7) The step of intercepting the discovery or request message that is part of the configuration dialogue between the configuration server and the client, and examining the discovery or request message to see if the broadcast flag for that message is set. And the step of setting the broadcast flag if the broadcast flag is not set, recalculating the checksum associated with the discovery message or request message, and transmitting the discovery message or request message. The client configuration method according to (1) above, further including.

【0084】

(8) The client configuration method according to (1) above, further including a step of storing the network address of the configuration agent in the configuration server address field of the above message.

【0085】

(9) A step of determining whether or not the above message is an acknowledgment message, and a step of storing one entry for associating the client with the above configuration server in the server allocation table if the message is an acknowledgment message. The client configuration method according to (1) above, including.

【0086】

(10) A step of intercepting a request message containing a network address requesting the client to renew its lease, a step of readdressing the request message to the address of the configuration server, and a step of transmitting the request message to the configuration server. The client configuration method according to (1) above, further including.

【0087】

(11) Determine whether the configuration agent and the configuration server are on the same subnet, and if the configuration agent and the configuration server are on the same subnet, the broadcast flag if the broadcast flag of the above request message is not set. The above step of readdressing the request message to the address of the configuration server includes the step of recalculating the checksum associated with the above request message, and the above step is to set the hardware address of the above request message to the hardware of the above request message. -Including the step of readdressing the address, if the configuration agent and the configuration server are not on the same subnet, the broadcast flag of the above request message is released, and the relay agent of the above request message is released. The above step, which includes the step of storing the configuration agent's network address in the field and recalculating the checksum associated with the above request message, and readdressing the request message to the configuration server's address, is the configuration server's network. The client configuration method according to (10) above, which includes a step of readdressing the network address of the above request message as an address.

【0088】

(12) The above step of readdressing the above request message to the address of the configuration server accesses the server allocation table, retrieves the address of the configuration server already associated with the client, and configures the configuration already associated with the client. The client configuration method according to (10) above, which includes a step of readdressing the above request message to the server address.

【0089】

(13) The step of intercepting discovery or request messages that are part of the configuration dialogue between the configuration server and the client, the step of determining if the configuration agent and the configuration server are on the same subnet, and the configuration agent and If the configuration server is on the same subnet and the broadcast flag is not set, set the broadcast flag for the discover message or request message and recalculate the checksum associated with the discovery message or request message. If the step of transmitting the discovery message or the request message and the configuration agent and the configuration server are not on the same subnet, the broadcast flag of the discovery message or request message is cleared if the broadcast flag is not cleared, and the discovery message is cleared. Alternatively, the relay agent field of the request message contains the steps of storing the network address of the configuration agent, recalculating the discovery message or the checksum associated with the request message, and transmitting the discovery message or request message, as described above. The client configuration method described in (1).

【0090】

(14) Configuration where the client can be configured with the default gateway address The first port connected to the server and the second port connected to the client requesting to be configured with the default gateway address A network device including a port, a sending device connected between the first port and the second port, and sending a packet between the first and second ports, wherein the sending device is , Includes a configuration agent that intercepts configuration packets that are part of the configuration dialogue between the configuration server and the client, the configuration packet contains the default gateway address used by the client, and the configuration agent is the client's. A network device that modifies the default gateway address so that it is a network address and transmits the configuration packet.

【0091】

(15) The network device according to (14) above, wherein the configuration packet includes an offer packet.

【0092】

(16) The network device according to (14) above, wherein the configuration packet includes an acknowledgment packet.

【0093】

(17) The configuration agent intercepts the configuration packet by receiving the configuration packet from the configuration server on the first port, and transmits the configuration packet to the client on the second port. The network device according to (14) above, which transmits packets.

【0094】

(18) Copy the candidate network address currently used or used as the network address of the client from the client network address field of the configuration packet, and the default gateway address of the configuration packet. The network device according to (14) above, wherein the configuration agent modifies the default gateway address to be used by the client by storing the candidate address in a field.

【0095】

(19) The network device according to (14) above, wherein the configuration agent recalculates the checksum associated with the configuration packet before transmitting the configuration packet.

【0096】

(20) The configuration agent intercepts a packet containing a discovery message or request message that is part of a configuration dialogue between the configuration server and the client, inspects the discovery message or request message, and is flagged as broadcast. The network device according to (14) above, which sets the broadcast flag of the discovery message or the request message, recalculates the checksum associated with the packet, and transmits the request packet, if not set.

【0097】

(21) The network device according to (14) above, wherein the configuration agent stores the network address of the configuration agent in the configuration server address field of the configuration packet.

【0098】

(22) The configuration agent determines whether or not the configuration packet contains an acknowledgment message, and if the configuration packet contains an acknowledgment message, an entry for associating the client with the configuration server is added to the server allocation table. The network device according to (14) above, which is stored.

【0099】

(23) The configuration agent intercepts the packet containing the request message containing the network address requesting the client to renew its lease, readdresses the packet to the address of the configuration server, and configures the packet. The network device according to (14) above, which is transmitted to a server.

【0100】

(24) The above configuration agent determines whether the configuration agent and the configuration server are on the same subnet, and if the configuration agent and the configuration server are on the same subnet, the broadcast flag of the above request message is set. If not, set the broadcast flag, recalculate the checksum associated with the packet, readdress the packet to the network address of the configuration server, and if the configuration agent and the configuration server are not on the same subnet, the above If the broadcast flag of the request message is not cleared, the broadcast flag is cleared, the network address of the configuration agent is stored in the relay agent field of the request message, and the checksum associated with the request message is recalculated. , The network device according to (23) above, which readdresses the packet to the network address of the configuration server.

【0101】

(25) The above configuration agent accesses the configuration server allocation table, retrieves the address of the configuration server already associated with the client, and readdresses the above packet to the address of the configuration server already associated with the client. The network device according to (23) above, which readdresses the packet to the address of the configuration server.

【0102】

(26) The above configuration agent intercepts the packet containing the discovery message or request message that is part of the configuration dialogue between the configuration agent and the client to determine whether the configuration agent and the configuration server are on the same subnet. However, if the configuration agent and the configuration server are on the same subnet, if the broadcast flag is not set, set the broadcast flag of the discovery message or request message, recalculate the checksum associated with the above packet, and recalculate the checksum associated with the above packet. If the above packet is transmitted and the configuration agent and the configuration server are not on the same subnet, the broadcast flag of the discovery message or request message is released if the broadcast flag is not cleared, and the relay agent of the discovery message or request message is released. -The network device according to (14) above, which stores the network address of the configuration agent in the field, recalculates the checksum related to the packet, and transmits the packet.

【0103】

[Effect of the invention]

By allowing a network node to be automatically configured as its own gateway, the present invention maintains automatic configuration and without significantly increasing traffic through the router. Allow network administrators to extend the network by adding switches and bridges. Because switches and bridges are much faster and cheaper than routers, the present invention gives network administrators the opportunity to maintain automatic node configuration while minimizing the cost of extending the network and maximizing the speed of the network. provide.

[Simple explanation of drawings]

[Figure 1]

It is a block diagram of the OSI standard model.

[Figure 2]

It is a block diagram which shows the network of the prior art.

[Fig. 3]

It is a block diagram which shows the network including the switch which has the configuration agent according to this invention.

[Fig. 4]

FIG. 5 is a block diagram showing an N-port switch according to a first embodiment of the present invention in which the configuration agent is a DHCP snooping agent.

[Fig. 5]

It is a flow chart which shows the method of processing a packet by the DHCP snooping agent shown in FIG.

[Fig. 6]

FIG. 5 is a block diagram showing an N-port switch according to a second embodiment of the present invention in which the configuration agent is a DHCP surrogate agent.

[Fig. 7]

Along with FIGS. 8 to 10, it is a flow chart showing a method of processing a packet by the DHCP proxy agent of FIG.

[Fig. 8]

Along with FIGS. 7, 9 and 10, it is a flow chart showing a method of processing a packet by the DHCP proxy agent of FIG.

[Fig. 9]

Along with FIGS. 7, 8 and 10, it is a flow chart showing a method of processing a packet by the DHCP proxy agent of FIG.

[Fig. 10]

Along with FIGS. 7 to 9, it is a flow chart showing a method of processing a packet by the DHCP proxy agent of FIG.

[Explanation of symbols]

74 network 76 Switch containing configuration agent 86 Network nodes to be configured (clients) 88, 132 DHCP server 100, 120 transmitter 102 DHCP snooping agent 130 DHCP proxy agent 134 DHCP / BOOTP relay snooping agent