Patent Application
20050094636
This application claims the priority of Korean Patent Application No. 2003-78127, filed on Nov. 5, 2003, in the Korean Intellectual Property Office, the disclosure of which is incorporated herein in its entirety by reference.
1. Field of the Invention
The present invention relates to a multi-protocol label switching (MPLS)-based packet switching apparatus and method, and more particularly, to a MPLS apparatus and method for guaranteeing continuity of services and reducing network loads.
2. Description of the Related Art
Multi-protocol label switching (MPLS) is a standards-approved technology for speeding up network traffic flows and making it easier to manage networks. MPLS involves setting up a specific path for a given sequence of packets, each of which is identified by a label. Each router searches for an address of a node, to which each of the packets is to be forwarded, by referring to the label of each of the packets, thereby saving the time required for each router to search for the address of the node, to which each of the packet is to be forward. This technology is called multi-protocol label switching rather than, for example, single protocol label switching, because it works with Internet protocol (IP), asynchronous transport mode (ATM) protocol, and frame relay network protocol. MPLS transmits packets mostly on Layer 2 of the OSI standard reference model rather than Layer 3 of the OSI standard reference model. MPLS makes it easier to manage a network for quality of service (QoS) as well as enhances the speed of network traffic flows. Until now, various efforts have been made to establish MPLS networks and provide distinctive MPLS services with a high QoS by using various signaling protocols, such as Label Distribution Protocol (LDP), Constraint-based Routing LDP (CR-LDP), and Resource reSerVation Protocol-Traffic Engineering (RSVP-TE). Additionally, various techniques of setting up Label Switched Paths (LSPs) with different reliabilities for different classes of services, protecting the LSPs, and recovering the LSPs if the LSPs are damaged have been suggested.
The above-described efforts to provide high-quality MPLS network services, however, may increase the number of control messages directed to each node of a network or the network itself and may eventually hinder the network from providing normal services by burdening the network with loads of control work. Conventional MPLS network services need periodic control messages for setting up and maintaining regular LSPs. Thus, in some cases, a conventional MPLS network may not be able to provide data forwarding services due to control plane errors even when continuity of the LSPs is guaranteed, i.e., even when the regular LSPs, which are data paths, are available for transmitting services therealong. In addition, in the case of protecting or recovering the LSPs in order to guarantee the QoS, reliability and continuity of the LSPs, it is necessary to use more complicated operations and to more frequently issue control messages, which results in a considerable increase in the workload of each node of the conventional MPLS network or the conventional MPLS network itself and data traffic loss until the LSPs are recovered.
The present invention provides a multi protocol label switching (MPLS) apparatus and method that can reduce node or network loads and guarantee the continuity of services by minimizing transmissions of control messages required for establishing and maintaining label switched paths (LSPs) and transmitting packets using an IP forwarding technique when the LSPs malfunction.
According to an aspect of the present invention, there is provided a method of establishing a label switched path (LSP) in a multi protocol label switching (MPLS) system. The method involves (a) setting an economy class service-forwarding equivalence class (ES-FEC), the ES-FEC supporting both an LSP forwarding method and an IP forwarding; (b) determining whether a packet satisfies the ES-FEC based on a header of the packet; (c) if the packet satisfies the ES-FEC, generating a first path establishment message containing identification data and transmitting the first path establishment message, the identification data being used for identifying the ES-FEC; and (d) establishing an LSP corresponding to the ES-FEC based on a second path establishment message and forming a forwarding table comprising a field designating either the LSP forwarding method or the IP forwarding method depending on whether an error has occurred at the LSP, the second path establishment message being received from nodes of an MPLS network in response to the first path establishment message.
According to another aspect of the present invention, there is provided a method of forwarding a packet in a MPLS system. The method involves (a) if an error has occurred at an LSP between one node to another node, recording information indicating that an error has occurred at the LSP in an error information field of a forwarding table; (b) reading an error information field corresponding to an input packet from the forwarding table; and (c) selecting a first forwarding method, in which the input packet is forwarded based on a label, or a second forwarding method, in which the input packet is forwarded based on a destination address of the input packet, with reference to the error information field read in (b).
According to another aspect of the present invention, there is provided an fault management method for an MPLS system. The falut management method involves (a) if an error has occurred at an LSP between one node and another node transmitting an error recovery sensing message; and (b) if an error recovery message is received in response to the error recovery sensing message, deleting information indicating that an error has occurred at the LSP from an error information field of a forwarding table.
According to another aspect of the present invention, there is provided an MPLS apparatus. The MPLS apparatus includes an ES-FEC setting unit, which sets an economy class service-forwarding equivalence class (ES-FEC), the ES-FEC supporting both an LSP forwarding method and an IP forwarding; a message management unit, which generates a first path establishment message containing identification data and transmits the first path establishment message, if a packet is determined to satisfy the ES-FEC based on a header of the packet; and a forwarding table management unit, which establishes an LSP corresponding to the ES-FEC based on a second path establishment message and forms a forwarding table comprising a field designating either the LSP forwarding method or the IP forwarding method depending on whether an error has occurred at the LSP, the second path establishment message being received from nodes of an MPLS network in response to the first path establishment message.
According to another aspect of the present invention, there is provided an MPLS apparatus. The MPLS apparatus includes an error setting unit, which, if an error has occurred at an LSP, records information indicating that an error has occurred at the LSP in an error information field of a forwarding table and IP-forwards an input packet based on a destination address of the input packet; and an error recovery unit, which transmits an error recovery sensing message if an error has occurred at the LSP, and deletes the information from the error information field of the forwarding table and then LSP-forwards the input packet if it receives an error recovery message in response to the error recovery sensing message.
Accordingly, it is possible to reduce node or network loads and guarantee continuity of services even when some LSPs in a network are blocked.
The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings in which:
The present invention will now be described more fully with reference to the accompanying drawings, in which exemplary embodiments of the invention are shown.
The destination address field 101 contains an address of a destination of each packet and the next hop address field 102 contains an address of a next node, to which each packet is transmitted.
The input label field 103 contains an input label, which is included in an input packet and is used for forwarding the input packet in a multi protocol label switching (MPLS) network. Each node searches the input label field 103 of a corresponding forwarding table for the input label and checks the output label field 104 of the corresponding forwarding table, which is provided in association with the input label field 103. The output label field 104 contains an output label that specifies a predetermined node, to which each input packet is to be forwarded. Once each input packet is forwarded to the predetermined node, label information included in the input packet is replaced with the value of the output label field 104.
The ES-LSP field 105 is a field that indicates whether an error has occurred at a Label Switched Path (LSP) in the MPLS network. The LSP may be set up by using a signalling protocol, such as Label Distribution Protocol (LDP, RFC 3036), Constraint-based Routing Label Distribution Protocol (CR-LDP), or Resource reSerVation Protocol-Traffic Engineering (RSVP-TE). If an error has occurred at an LSP, ‘BLOCKED’ is written in the ES-LSP field 105. Otherwise, ‘UNBLOCKED’ is written in the ES-LSP field 105.
The ES-LSP field 105 is introduced for providing economy-class services. Here, the economy-class services denote services provided while reducing node and/or network loads and guaranteeing service continuity. The ES-LSP field 105 can be divided into two session fields specifying whether an error has occurred at an upstream session or a downstream session of the LSP or both. For example, if an error has occurred at an LSP 160 between the nodes B and C, ‘BLOCKED’ is written in an upstream session field of an ES-LSP field of the forwarding table 110 for the node B, and ‘UNBLOCKED’ is written in a downstream session field of the ES-LSP field of the forwarding table 110 for the node B. The ES-LSP field 105 will be described more fully with reference to
The nodes A, B, C, and D each include a session table (140 or 150) as well as their respective forwarding tables 100, 110, 120, and 130. The session tables 140 and 150 include upstream session fields 142 and 152, respectively, and downstream session fields 144 and 154, respectively, and store upstream session information, downstream session information, and the frequency of transmitting a control message.
The session table 140 for the node B includes an ES-LSP field in order to support ES-LSPs according to the present invention. The nodes A, B, C, and D periodically transmit a message to one another to check whether LSPs thereamong are blocked. In the case of using LDP, the nodes A, B, C, and D periodically transmit a ‘KeepAlive’ message to one another to check whether the LDPs thereamong are blocked, in which case, the frequency of transmitting the ‘KeepAlive’ message among the nodes A, B, C, and D is stored in the session table for each of the nodes A, B, C, and D. Preferably, but not necessarily, the frequency of transmitting a predetermined control message, such as the ‘KeepAlive’ message, among the nodes A, B, C, and D is set low.
A MPLS apparatus and method according to the present invention maintain connections among nodes even when some of the LSPs provided thereamong are blocked, by performing an IP forwarding process. Therefore, even when the frequency of transmitting a control message among the nodes for sensing whether the LSPs thereamong are blocked is set low, it is possible to guarantee continuity and quality-of-service (QoS) of services.
It is assumed that a path A-B-C-D in
An ingress LER determines an LSP, along which a packet received from a non-MPLS network is to be forwarded, by referring to a destination IP address contained in a header of the packet. When a packet with a label is received from an LER, an LSR only checks the label, resets a value of the label of the packet, and forwards the packet to a predetermined outgoing interface. The ingress LER encapsulates a packet according to an outgoing interface (Layer 2) of the packet. An egress LER removes a label from a packet received from an MPLS network and forwards the packet to a predetermined destination specified in the packet. In the MPLS network, the LER, which is a terminal node of an LSP, performs a Layer 3 packet forwarding process, and the LSR, which is an inbetween node of the LSP, performs a Layer 2 forwarding process. Each node in the MPLS network generates a label information table and a forwarding table by using LDP.
If an error has occurred at the LSP 160 between the nodes B and C, ‘BLOCKED’ is written in the ES-LSP field 105 of each of the forwarding tables 110 and 120. The node B searches the forwarding table 120 for an entry corresponding to a packet input thereto and forwards the input packet to a subsequent node by using an MPLS method, if ‘UNBLOCKED’ is written in the ES-LSP field of the searched entry of the forwarding table 110. In other words, the node B switches a label of the input packet with an output label of the forwarding table 110 and then forwards the input packet to the subsequent node.
If ‘BLOCKED’ is written in the ES-LSP field of the searched entry of the forwarding table 110, the node B forwards the input packet to the subsequent node by using an IP forwarding method. Referring back to
The node C searches the forwarding table 120 for an entry corresponding to the packet received from the node E. The node C adds a label to the received packet and then forwards the received packet to the node D along an LSP therebetween. Accordingly, even though the LSP 160 between the nodes B and C is blocked, it is possible to guarantee of QoS of a service by guaranteeing the continuity of the transmission of the packet from the node A to the node D.
A general FEC TLV and an FEC TLV for an ES LSP according to the present invention have the same data structure. In order to differentiate the FEC TLV for an ES LSP from the general FEC TLV, a specific value is assigned to the FEC TLV for an ES LSP, which will be described more fully with reference to
The node transmits the general FEC TLV or the FEC TLV for an ES LSP to the outside in operation S215 or S260. The node receives a general FEC TLV or an FEC TLV for an ES LSP in operation S215 or S275. If the node adopts an RFC 3036 downstream-on-demand (DoD) manner, it transmits a label request message to a neighboring node. If the node adopts an RFC 3036 downstream unsolicited (DU) manner, it transmits a label mapping message to the neighboring node. Here, DoD and DU are different types of label distribution manners.
More specifically, DoD distributes a label on a “on demand” basis. Accordingly, a downstream node that receives a label request message sends a label mapping message containing label information to an upstream node. On the other hand, DU distributes a label in an unsolicited manner. Accordingly, once the downstream node recognizes FEC information, it sends the label mapping message to the upstream node.
Various types of messages can be transmitted between the downstream and upstream nodes, and the sequence of the messages transmitted between the downstream and upstream nodes may vary. However, all of the messages transmitted between the downstream and upstream nodes contain either a general FEC TLV or an FEC TLV for an ES LSP.
In operation S280, the node checks whether the received FEC TLV is a general FEC TLV or an FEC TLV for an ES LSP. If the received FEC TLV is an FEC TLV for an ES LSP, the node forms a forwarding table with an LSP field, which specifies whether an LSP is blocked, in operation S215 or S285. If the received FEC TLV is not an FEC TLV for an ES LSP, the node records ‘NONE’ in the LSP field in operation S215 or S290.
Session tables 340 and 350 and forwarding tables 300, 310, 320, and 330 are the same as their respective counterparts of
If the LSP 360 is blocked between the nodes B and C, ‘DOWNSTREAM BLOCKED’ is written in an ES-LSP field 312 of the forwarding table 310 for the node B, ‘UPSTREAM BLOCKED’ is written in the ES-LSP field 322 of the forwarding table 320 for the node C, ‘BLOCKED’ is written in a downstream session field of the session table 340 for the node B 380, and ‘BLOCKED’ is written in an upstream session field 352 of the session table 350 for the node C 390.
DOWNSTREAM denotes a downstream session for a packet, and UPSTREAM denotes an upstream session for the packet. BLOCKED denotes that a corresponding section of an LSP is blocked, and UNBLOCKED denotes that the corresponding section of the LSP is unblocked. Accordingly, it is possible to efficiently represent whether each of the upstream and downstream sessions for the packet is blocked, by using any of “BLOCKED” or “UNBLOCKED”, or “DOWNSTREAM BLOCKED” or “UPSTREAM BLOCKED”.
As shown in the session table 340 for the node B, ‘BLOCKED’ is written in the downstream session field 344, but ‘UNBLOCKED’ is written in the upstream session field 342. Accordingly, when the node B receives a packet, it checks the forwarding table 320 and recognizes that a downstream session is blocked. Then, the node B removes a label from the packet and then forwards the packet by using an IP forwarding method rather than a typical LSP method. In other words, the node B forwards the packet to a node E rather than to the node C along the IP forwarding path 370. Then, the node E IP-forwards the packet received from the node B to the node C.
When the node C receives the packet from the node E, the node C checks the forwarding table 310 and recognizes that an upstream session is blocked but a downstream session is unblocked. Thereafter, if the packet received from the node E has been IP-forwarded from the node E, the node C forwards adds a label to the packet and forwards the packet to the node D along the LSP 360. If the node D is an egress LER in an MPLS network, it removes the label from the packet and then IP-forwards the packet to a destination of the packet. In short, the packet is forwarded from the node A to the node B, from the node B to the node E, from the node E to the node C, and from the node C to the node D. In particular, an IP forwarding method is used for forwarding the packet from the node B to the node E and from the node E to the node C, and a typical LSP method is used for forwarding the packet from the node A to the node B and from the node C to the node D.
As described above, if a section of an LSP is erroneous, the section is blocked, and the rest of the LSP is maintained to function normally. Then, packets are forwarded along the LSP bypassing the error-occurred section of the LSP through IP forwarding, thereby guaranteeing continuity of the transmission of the packets.
In the present invention, if an error has occurred at a section of an LSP during a packet forwarding service, nodes in the vicinity of the predetermined session stop operating a ‘KeepAlive’ timer without transmitting a notification message or a label release/withdrawal message according to the RFC 3036. Therefore, the nodes of the error-occurred section only transmit a Hello message indicating that they have sensed recovery of the error-occurred section, and other nodes process packets as if the LSP had never been blocked. In other words, nodes along the entire LSP but the error-occurred section consider the LSP as normal and perform the packet forwarding service in a label switching method, and the nodes of the error-occurred section perform the packet forwarding service in an IP forwarding method until the predetermined section is recovered.
If the error-occurred section is recovered, only the nodes of the error-occurred section recognize the recovery of the error-occurred section and takes advantage of LSP information that could have been abandoned when an error occurred to forward packets in a label switching method in the recovered section.
The predetermined node receives an error recovery message from the adjacent node in operation S455 and then performs appropriate processes, e.g., session setting. More specifically, when the predetermined node receives a Hello message, i.e., the error recovery message, from the adjacent node as a response to another Hello message that it has transmitted to the adjacent node, i.e., the error recovery sensing message, it performs appropriate processes, e.g., session setting.
The error recovery sensing message and the error recovery message may be transmitted between the predetermined node and the adjacent node when a daemon, which is a process that runs in the background. The predetermined node searches a session table based on an identifier of the adjacent node contained in the error recovery message in operation S460. The session table stores session information for management of a session for any two nodes, and the interval of transmission of a KeepAlive message.
If an entry corresponding to the error recovery message received from the adjacent node exists in the session table, and “BLOCKED” is recorded in an ES-LSP field of the entry in operation S654, the predetermined node records “UNBLOCKED” in an ES-LSP field of the session table and in an ES-LSP field of a forwarding table. Once the LSP between the predetermined node and the adjacent node is recovered, previously set session information can be used without setting a new session for the predetermined node and the adjacent node. If the entry corresponding to the identifier of the adjacent node does not exist in the session table, the predetermined node sets a session in operation S475 by performing a predetermined message exchange process.
UNBLOCKED 520 denotes a normal state, in which an error is yet to occur at the LSP, as indicated by 525 of
ES-Prefix 612 indicates that a packet is classified into one of a plurality of FEC groups based on a prefix of a destination address of the packet and that the FEC of the packet supports the ES-LSP according to the present invention. In order to differentiate the FEC of the packet from other general FECs that do not support the ES-LSP according to the present invention, ES-Prefix 612 generally uses a type value of 0×16. Address Family 614 generally uses values defined in the RFC 1700, and each of the values indicates which address family among IP, IPv6, NSAP, E.163, and so forth an address included in the FEC TLV belongs to.
PreLen 616 specifies the length of Prefix 618. Prefix 618 indicates which class an IP address of a network belongs to. ES-HostADDr 622 indicates that the FEC of the packet can be identified based on a destination host address of the packet and that the FEC of the packet supports the ES-LSP according to the present invention. In order to differentiate the FEC of the packet from the other general FECs that do not support the ES-LSP according to the present invention, ES-HostAddr 622 uses a type value of 0×17. Address Family 624 is the same as Address Family 614. HostAddr Len 626 specifies the length of the host address of the packet. Host Addr 628 specifies the host address of the packet.
As described above with reference to
The message management unit 710 identifies an FEC of an input packet based on a header of the input packet. If the FEC of the input packet is an ES-FEC, a path establishment message containing identification data for identifying the ES-FEC is generated and transmitted. The path establishment message includes label information and is generated as the FEC TLV of
The table management unit 720 receives a path establishment message from the outside of the MPLS apparatus and forms a forwarding table which comprises a field specifying whether the LSP forwarding method or the IP forwarding method is to be used to forward the input packet. Whether the input packet is to be forwarded by using the LSP forwarding method or the IP forwarding method is determined based on whether an error has occurred at an LSP, along which the input packet is to be forwarded. The table management unit 720 records predetermined data in the field of the forwarding table so that if an error has occurred at the LSP, the input packet can be IP-forwarded, and once the error is recovered, the input packet can be LSP-forwarded. The type of the predetermined data has already been described above with reference to
Therefore, once the formation of the forwarding table is completed, and a packet is received, a node searches for a next destination of the packet with reference to the forwarding table and then forwards the packet to the searched destination.
If an error has occurred at an LSP, the error setting unit 750 records information indicating an error has occurred at the LSP in an error information field of a forwarding table and forwards an input packet based on a destination address of the input packet.
The error recovery unit 760 transmits an error recovery sensing message. If the error recovery unit 760 receives an error recovery message in response to the error recovery sensing message, it deletes the error information from the forwarding table and then LSP-forwards the input packet.
The present invention can be realized as a computer-readable code written on a computer-readable recording medium. The computer-readable recording medium includes nearly all kinds of recording devices, on which computer-readable data can be recorded. For example, the computer-readable recording medium includes a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage, and a carrier wave (e.g., data transmission through the Internet). In addition, the computer-readable recording medium can be distributed over a plurality of computer systems connected to a network so that a computer-readable code written thereon can be executed in a decentralized manner.
According to the present invention, when an error has occurred at an LSP, a packet is IP-forwarded. Once the error is recovered, a session is recovered by using existing session information, and then the packet is LSP-forwarded. Therefore, it is possible to guarantee the continuity and QoS of packet services.
In addition, even when an error has occurred at the LSP, it is possible to lengthen the cycle of transmitting a control message to maintain the LSP or detect the error by IP-forwarding the packet. Accordingly, it is possible to considerably reduce network/node loads.
Moreover, nodes adjacent to the LSP, at which an error has occurred, can use an existing session without the need to set up a new session, thus considerably reducing session loads.
While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2003-78127 | Nov 2003 | KR | national |
