Method for port connectivity discovery in transparent high bandwidth networks

[0002] N/A

[0003] N/A

[0004] The present invention relates generally to high bandwidth networks and methods used for managing such networks and, more specifically, to optical networks and methods used for discovering the connectivity of networks.

[0005] One traditional way of tracking the the connectivity of communication links in a network has been to record the connectivity manually as the node elements and links are installed and activated. Further updates to the network might not be recorded resulting in the record of the circuit-based network becoming obsolete. As packet-based communications became the primary communications protocol to utilize these networks, computer-controlled management of the packets in network was developed and utilized to track packets. However, similar network management methods were not developed for the circuit-based underlying structure of the network. Networks that use a protocol that allows for neighbor and connectivity discovery are termed auto-discovering. Packet-based networks, such as TCP/IP and OSI networks use this technique.

[0006] In circuit switched networks, however, connections have traditionally been hand-wired. Circuit based protocols were not designed for auto-discovery. While some capability might be designed into the newer optical connection products, the standards for circuit-based networks are currently not sufficiently developed to allow self-discovery of a general circuit-based network. Recently, data bandwidth requirements have increased and the new types of data (digitized video, massive databases, etc) are too fast to be served by packet-based protocols.

[0007] High bandwidth networks require entire circuits, the communication connection between two points, to carry data. The need to understand the topology of the layout of circuits is too critical to rely on the unreliable records of the network generated by yesterday's technology. Therefore, there is a need for a methodology of mapping the circuits present in such networks.

[0008] A method of determining the connectivity of a circuit-based network uses the information currently available about the network; its nodes, the ports on the nodes, and the type of the links between ports, to support a technique to discover the remaining connectivity information about the network. A management entity controls the discovery process by enabling transmission of a unique stylized probe message (a TRACE message in the described embodiments) containing transmitter specific information. When a receiver recognizes the message, it reports both the receiver's identity and the transmitter's identity to the management entity. By accumulating this data, the management entity accumulates the connectivity information about the network. Each probe message yields at most one link in the connectivity pattern. The unique stylized probe messages are tailored to the transmission mechanism available at each transmitter. While different types of transmission mechanisms can be used, it is preferable to use the one capable of carrying the most information and disrupting the network least. The management entity that controls the mapping operation can be monolithic, building a map of the entire network in one database, or distributed, allowing maps of subnetworks to reside in multiple nodes hosting management components and one coordinating management resource able to access all the connection information.

[0009] The invention will be more fully understood by reference to the following detailed description when considered in conjunction with the accompanying drawings, in which:

[0010] FIG. 1 is a schematic of a network according to the invention;

[0011] FIG. 2 is a detail of two ports from the network of FIG. 1, and their connection as a circuit;

[0012] FIG. 3 is a block diagram illustrating participating ports and non-participating ports according to the invention;

[0013] FIG. 4 is a state diagram of a port according to the invention;

[0014] FIG. 5 is a representative list maintained by a management component in the network of FIG. 1, of its ports, their connections and states;

[0015] FIG. 6 is a flow diagram of the operation of a management entity discovering the connectivity of the network of FIG. 3;

[0016] FIG. 7 is a matrix showing the possible input/output characteristics of each side of a port of FIG. 2 and the state that corresponds with each; and

[0017] FIG. 8 is a state diagram of a managed port according to the invention.

[0018] The methods described herein are applicable to high speed switched networks. Examples from an optical implementation will be used to illustrate such methods.

[0019] The network illustrated in FIG. 1 shows two managed circuit based networks 10, 14 that have network-to-network connections traversing a transparent switched network 12. The networks 10 and 14 are connected to respective sets of users via links 18 and 20. The networks 10 and 14 are further connected to the transparent switched network 12 by network to network connections 16.

[0020] The managed circuit switched network 10, includes a management entity (not shown) that tracks the state of the network. The state of the network includes information about physical links interconnecting ports, which ports are being used by a program or user, and which ports are not linked to another port. The transparent switched network 12 is not part of the managed switched network 10, so the management entity in managed switched network 10 doesn't know the state of the connections in the transparent switched network 12. If the elements of FIG. 1 are part of a larger managed network, a coordinating management resource (not shown) needs to know the connections that pass through the transparent switched network 12. However, the management entity in the managed switched network 10 generally does not know the connectivity within the transparent network 12, nor where connections 16 terminate in the transparent network 12. Additionally, in transparent networks it is not possible to modify the traffic stream in order to determine connectivity.

[0021] FIG. 2 shows a detail of two interconnected ports in the network. Bi-directional ports are represented by 30 and 32. The bi-directional port 30 incorporates two sides, a transmitter 38 transmitting information from A to B, and a receiver 34 receiving information transmitted from B to A. The link 31 may consist of one fiber carrying signals in both directions, or a pair of fibers, each carrying one unidirectional signal.

[0022] A receiver such as 34 is generally always active. It is monitoring its input. The receiver 34 either receives nothing or it receives information in a form it can interpret. Thus, a receiver connected to a management entity can inform the management entity whether it is connected to an active transmitter. If the transmitter 36 sends a unique message, called a TRACE message, then receiver 34 can detect this message and report it to the management entity.

[0023] The mechanism for carrying a TRACE message can be other than a normal information carrying mechanism; it could be a unique frequency, a variation in voltage, or some other pre-arranged mechanism. The TRACE message also carries unique information such as an identifying address which can be passed on to the management entity allowing the management entity to infer the existence of a circuit between the reporting receiver 34 and the transmitter 36 sending the unique message.

[0024] The transmitter 36, on the other hand, is passive. It merely transmits a signal presented to it irrespective of the presence of a receiver. The transmitter can send a unique signal, but cannot detect whether it is received nor identify the receiver.

[0025] FIG. 3 illustrates the communications between ports and management components including coordinating management resources. A first set of ports 63, 64, and 65 report to one management component 60. A second set of ports 66, 67, and 68 report to another management component 58. Because of the connections between ports, management component 60 learns of ports 67 and 68, and management component 58 learns of ports 63 and 64. In addition, management component 60, due to its connection with port 65, learns of port 69, which does not report to any management component. Ports 54 and 56 are connected to each other, but do not participate in the managed network because neither one of them reports to a management entity. The coordinating management component 62 may maintain an overall database of ports in the entire managed network and/or may facilitate the transfer of information between all management resources including ones not shown.

[0026] The management entity tracks the status of each port using a sequence of states as illustrated in FIG. 4. A port can be either connected to another port or disconnected. Line 71 separates the port state diagram into a connected side containing state 80 and disconnected side containing states 70 and 74. A port starts in state NOT-CONNECTED 70 meaning that the port can report to the management entity but is not transmitting to nor receiving from any other port. A management command 72 moves the port into the TRACING state 74 meaning that the transmitter side of the port is sending a unique message (the TRACE message) that, if received and reported to the management entity by a managed receiver, allows the management entity to infer a link between the two ports. The port remains in the TRACING state 74 until the port's TRACE message has been received by a receiver whereupon a management command 78 moves the port into the CONNECTED state 80. When the port is in CONNECTED state 80, it is available for use by communications applications programs. If the port's TRACE message is not reported as received within a predetermined time, the management entity sends a command 75 that causes the port to transition from the TRACING state 74 to the NOT-CONNECTED state 70. When a port has entered the CONNECTED state 80, it remains in the CONNECTED state 80 until the connection to its counterpart is lost. When the management entity is informed of the lost connection, it sends a command 82 that moves the port to the NOT-CONNECTED state 70.

[0027] The management entity tracks port connectivity and tracing activity using a port list such as illustrated in FIG. 5, which shows a port list in the Coordinating Management Component 62 of FIG. 3. FIG. 5aillustrates an initial state of the list, for example upon power-up of the management unit or power-up of the entire network. The management entity knows of a number of ports by their ID, and for some of those ports knows the physical layer protocol or physical connection type. For instance in FIG. 5a, Port 63 is known to use the SONET protocol and Port 66 to use Ethernet. Techniques to determine the type of connection are well known in the art. It is assumed that appropriate techniques are used to determine the connection types for the remaining ports, and these values are then entered in the port list as shown in FIG. 5b.

[0028] In an alternate embodiment, if the management entity has a record of the interconnection of ports in the network from some source, such as manual records, the management entity could provisionally fill in the “Connected to” column of the list. These initial connections would be subject to verification by the TRACE mechanism, but would provide connection information for unmanaged ports. The verification would update the manual records for broken links, unrecorded upgrades or faulty equipment.

[0029] In an embodiment of the invention, initialization of the management port list of FIG. 5 and the network is tracked as shown in FIGS. 5athrough 5dand takes the form of the steps in the flowchart of FIG. 6. The management entity first establishes a reporting relationship with each port that it will be managing and receives status from each port, step 100. At the end of this operation, the port list has been updated as shown in FIG. 5b, in which ports 63 and 68 are shown as connected to ports 68 and 63 respectively and port 66 is shown as connected, but to an unknown port. Port 66 could be, for instance, connected to a user device.

[0030] Connectivity is then discovered using a path tracing technique that employs TRACE messages as mentioned above. A TRACE message is a unique message, distinguishable from a regular transmission, which carries information that can be interpreted by some receivers and reported back to a management entity. The report from the receiver allows the management entity to discover the link between the receiver and the transmitter that is sending the TRACE message.

[0031] As previously mentioned, the TRACE message may be carried over a link using any of a variety of TRACE mechanisms. The TRACE mechanisms may be stored in a database or a list and the “type of connection” entry in the port list of FIG. 5 provides a pointer for the management entity as to what type of TRACE mechanism to use. For example, a SONET link TRACE message could be sent using a pattern in the J0 (section trace) bytes. A Gigabit Ethernet link TRACE message could be sent using a standard packet or could use unused codewords of the 8B/10B code. For a transparent link capable of low frequency modulation and detection a TRACE message could be sent by varying the voltage of the signal. A TCP/IP or OSI link TRACE message could be sent using the self discovery capabilities built into the protocol. For a link where the signal can be turned on and off, and where the presence of a signal can be detected, a TRACE message could be sent by turning power on and off in a Morse Code like pattern. For a link capable of handling a subcarrier, such as one using some analog component before signal detection, a TRACE message could be sent by placing signals on a sub carrier. Finally, for a link where information is encoded into forward error correction frames, a TRACE message could be sent using Forward Error Correction Frames. This technique can be extended to any link where information is encoded into frames, and at least some of the bits are available when the frames are not carrying traffic. For some types of connection, multiple TRACE mechanisms could be possible, with the success of a particular mechanism dependent on the higher level protocols being used on the link. As an alternate embodiment, multiple parallel TRACE mechanisms may be initiated at one port, with the mechanism that successfully detects the connection being utilized for further TRACEs. As standards are developed in the optical networking realm, the management entity will be able to use the least intrusive and most informative standard supported by the link.

[0032] The TRACE message includes a TRACE signature to distinguish the TRACE message from another transmission, an identification of the transmitter port to allow the management entity to identify both ends of the link, and an identification of the management entity that needs to know the transmitter and receiver that define the link. The TRACE signature can be an unlikely constant value or some function of the rest of the data in the message. A type of checksum is one such function. The checksum can be further encoded by adding in a unique value to a message-computed checksum. A separate management component, such as the receiver's management component, may receive the initial TRACE report from the receiver, but that information will be passed through the distributed management to the identified management entity.

[0033] Referring to the flow diagram of FIG. 6, after the first contact with the ports is accomplished in step 100, the management entity directs a set of transmitters to send unique TRACE messages into the network, step 104. In a preferred embodiment, groups of transmitters source TRACE messages simultaneously, although for an initial setup situation, all disconnected ports could TRACE simultaneously. The management port list is further updated to the state of FIG. 5c, where port 65 is shown as TRACING. The management entity waits a predefined time for reports from receivers, step 106. Each time a report (referred to as a TRACE detected message) is received, the management port list is updated and the TRACE message(s) for that link are terminated, step 108. One report may cause termination of two TRACE messages, the TRACE message that caused the report by the receiving port and any TRACE message currently being sent by the transmitting side of the receiving port.

[0034] In a preferred embodiment, as a failsafe mechanism in case the management entity does not respond in a timely manner, the receiver of a port that has recognized a TRACE message terminates any TRACE message being sent by its own transmitter.

[0035] Once the predefined time has elapsed, an additional set of transmitters is directed to send TRACE messages, step 104 as shown in FIG. 5d, until all initially NOT-CONNECTED transmitters known to the management entity have been in the TRACING state. Once all disconnected ports have undergone a TRACE, the linked ports are freed to communicate in Step 110 by making the ports available to a communications program. Alternatively, linked ports are freed to communicate as soon as the management entity is sure that any TRACING of the line has been terminated. The management entity continues TRACING disconnected ports, step 112, according to a predetermined algorithm such as a round robin, while managing other aspects of the network. When all ports known to the management entity are connected, the management port list will look like FIG. 5e.

[0036] The management entity can serve functions other than the TRACING function. For example, for Ethernet links, the management entity can supply a port with its counterpart's IP address at the time the physical address of the link is reported.

[0037] The management entity presumes that links remain connected unless a communications program reports a disconnect. When a disconnect between two ports is reported, the management port list is updated to show that the ports are now NOT-CONNECTED, and a TRACE of that link is scheduled according to the predetermined algorithm.

[0038] An alternate way of viewing the state of a port is by the actions being performed by the two sides of the port as shown in FIG. 7. If the receiver is receiving data as shown in column 2 of FIG. 7, the port is in the CONNECTED state 80 regardless of what the transmitter is doing. The transmitter is not allowed to be sending a TRACE when its receiver is receiving. Similarly, if the receiver is receiving a valid carrier as shown in column 4 of FIG. 7, the port is in the CONNECTED state 80 and the transmitter is not allowed send a TRACE message. If the receiver has nothing on its input as shown in column 5 of FIG. 7, the port is disconnected. When the transmitter is TRACING while the receiver has nothing on its input, the port is in the TRACING state 74. If the transmitter is idle, the port is in the NOT-CONNECTED state 70. If the transmitter is sending data while the receiver has nothing on its input, the port is disconnected, but there is an error condition—likely a false positive on the transmitter. Similarly, if the receiver is receiving something other than data or a valid carrier as shown in column 3 of FIG. 7, the port is disconnected and can be in either the NOT-CONNECTED state 70 or TRACING state 74.

[0039] The method of using TRACE messages to discover connectivity can be extended to monitoring an active network. In this mode, the management entity uses an algorithm to choose which circuits to monitor. This algorithm can be based on a variety of information sources available to the management entity, such as user configuration, traffic type, incidence of false positive connections, and length of connection. The management port list is updated as shown in FIG. 5fto indicate that a port is being MONITORED. The TRACE message used for monitoring is a low-information-rate trace sent using a mechanism that is known not to be disruptive to the particular circuit. For example, information could be sent via the section trace byte in a SONET message by occasionally changing the byte.

[0040] The sequence of states employed for active monitoring is illustrated in the state diagram of FIG. 8. Line 118 separates the port state diagram into a connected side containing states 80, 136 and 154 and a disconnected side containing states 70 and 74. The disconnected side of the FIG. 8 is the same as illustrated in FIG. 4, while the connected side adds the MONITORED state 136 and the DISRUPTED state 154. When a connection to a port in the CONNECTED state 80 is lost, a management command 82 places the port in the NOT-CONNECTED state 70. If the CONNECTED port is to be monitored, a management command 134 puts the port into the MONITORED state 136 where monitoring traces are sent to the port by its counterpart. The port normally stays in the MONITORED state 136 until returned to the CONNECTED state 80 by a management command 140. If a disruption in the monitoring is detected, the port being MONITORED reports this to the management entity which issues a command 152 placing the port in the DISRUPTED state 154. If the disruption is cleared quickly enough, a later management command 150 will return the port to the MONITORED state 136. If the disruption is not cleared quickly, a management command 144 places the port in the NOT-CONNECTED state 70.

[0041] Preferred embodiments of the invention having been described, it will be apparent to those skilled in the art that other embodiments incorporating these concepts may be used. Accordingly, it is submitted that the invention should not be limited by the described embodiments but rather should only be limited by the spirit and scope of the appended claims.