JP2006229302A - Fault recovery system for l2 network at high speed - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a high speed fault recovery system stably recovering a fault in the L2 network at a high speed without inhibiting effective utilization of an interface of an L2 switch. <P>SOLUTION: Optical cross connect units (OXCs) 3, 4 are arranged between the L2 switches (L2SWs) 1, 2. When detecting an occurrence of a fault in a line (1), the L2SWs 1, 2 temporarily shift a traffic of the line whose fault is detected to another line (2) to which link aggregation (LAG) is set. When detecting the occurrence of a fault in the line (1), the OXCs 3, 4 switches the line to a restoration line using a standby line (3). Thereafter, the L2SWs 1, 2 carry out line switch-back using the LAG to restore the line by the LAG setting. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a high-speed failure recovery method for an L2 network, and more particularly to a high-speed failure recovery method that can recover a failure in an L2 network at high speed and stably.

  A wide-area L2 network configured by connecting a plurality of L2 switches (hereinafter referred to as L2SW) is described in Non-Patent Document 1, and STP (Spanning tree protocol) or RSTP is used as a failure recovery technique in such an L2 network. (Rapid STP) and LAG (Link aggregation) technologies are known. In STP and RSTP, in order to prevent broadcast packets from flowing infinitely through a loop, the loop is automatically detected and the loop is logically disconnected.

Also, the failure recovery mechanism using LAG performs traffic relief by avoiding the traffic on the line where the failure is detected by another line (backup line or shared line) set in advance by LAG. In this case, the traffic loss can be suppressed if the spare line preliminarily set with LAG is sufficient.
"10 Gigabit Ethernet textbook" published by IDG Japan

  However, in STP and RSTP, the time required for failure recovery is as long as several tens of seconds or several seconds, which causes unacceptable service interruption, and thus lacks reliability. Furthermore, STP and RSTP can automatically recover from a fault by performing topology conversion. However, there is no function for notifying which link a fault has occurred, and an explicit route according to the operator's policy. Operation monitoring and maintenance is hindered because selection is not possible.

  In addition, the LAG failure recovery mechanism can perform high-speed failure recovery. However, if there are only a few spare lines with LAG set in advance, traffic that cannot be relieved when a failure occurs in a high-traffic line. And traffic loss may occur. This traffic loss can be suppressed by setting a sufficient number of protection lines between the L2SWs. However, in this case, since the physical interface of the L2SW is occupied for the protection line, there arises a problem that the L2SW interface cannot be effectively used.

  An object of the present invention is to solve the above problems and provide a high-speed failure recovery method capable of recovering a failure in an L2 network at high speed and stably without hindering effective use of an L2SW interface.

  In order to solve the above problems, the present invention provides an optical cross-connect device (hereinafter referred to as OXC) between L2SWs, and when the L2SW detects a line failure, the traffic on the line where the failure is detected. Is temporarily transferred to another line set with LAG, and the line is restored by LAG setting after line switching by OXC. When the OXC detects a fault in the line, the OXC switches to the restoration line using the protection line. The first feature is that line switching is performed.

  In addition, the present invention has a second feature in that the OXC includes a timer control unit that delays the time from line fault detection to line switching.

  In addition, the present invention has a third feature in that a VLAN is allocated in units of a plurality of lines that are LAG-set by the L2SW.

  Furthermore, the present invention has a fourth feature in that the protection line is selected according to a line condition or an operator policy.

  According to the present invention, when a failure occurs in a line, the traffic is first relieved at a high speed by the LAG technology, and then the line is restored by line switching by OXC, and then the traffic that has been avoided by the LAG becomes a restoration line. Flowing. As a result, traffic loss can be minimized. In OXC, it becomes possible to explicitly select a route according to the line status and the operator's policy by the GMPLS protocol. Since there is no need to maintain a protection line between the L2SWs, the L2SW interface can be used effectively.

  Further, even if the LAG switching time and the OXC switching time are close to each other as they are, a stable failure recovery operation can be obtained. Further, it is possible to avoid the occurrence of traffic congestion without using STP or RSTP. Furthermore, traffic distribution can be achieved by explicitly selecting a protection line according to the line status or the operator's policy.

  The present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a basic configuration of an L2 network to which the present invention is applied. This L2 network has an L2SW1-OXC3-OXC4-L2SW2 configuration in which OXCs 3, 4 are arranged between L2 switches (hereinafter referred to as L2SW) 1,2.

  In FIG. 1, a line (1) between L2SWs 1 and 2 (indicated by a one-dot chain line) and a line (2) (indicated by a two-dot chain line) are set to LAG, and a VLAN 100 is assigned to the logical line configured thereby. Traffic (indicated by shading) flows in a distributed manner using a logical line composed of lines (1) and (2) as one used line. Congestion due to loop formation can be avoided by allocating a VLAN (Virtual LAN) for each logical line, that is, a plurality of lines (used lines) in which LAG is set between the L2SWs 1 and 2. In addition, there is a protection line (3) (indicated by a solid line) between the OXCs 3 and 4 apart from the used line, but the interface of the L2SWs 1 and 2 is not particularly occupied for the protection line (3).

  Next, a fast failure recovery method for the L2 network according to the present invention will be described. In the following, an embodiment when a failure occurs between the OXCs 3 and 4 of the line (1) of the L2 network configured as described above will be described. Note that the L2SW and OXC are arranged close to each other, and there is almost no failure between them.

  2 to 5 are diagrams sequentially illustrating operations from failure occurrence to failure recovery. First, when a failure (indicated by x) occurs in the line (1), the L2SWs 1 and 2 and the OXCs 3 and 4 detect this failure (FIG. 2). According to the GMPLS protocol, not only the occurrence of a failure but also the detection of the failure location can be performed, and this can be notified to the operator.

  The L2SWs 1 and 2 that have detected the failure temporarily transfer the traffic on the line (1) to the line (2) in which the LAG is set, and relieve it (FIG. 3). At this time, the traffic flowing through the line (2) temporarily increases. Since this transition is based on the LAG technology, it can be performed in a short time of about 50 ms. Depending on the available bandwidth of the line (2), there is a possibility that traffic that is not relieved here may occur, but this traffic is relieved by line switching by OXC later.

  The OXCs 3 and 4 that have detected the failure are switched so that the portion between the L2SW1 and OXC3 and the portion between the L2SW2 and OXC4 of the line (1) are connected to the protection line (3). That is, a recovery line using the protection line (3) is set (FIG. 4). It usually takes about 500 ms from the detection of a fault to this switching.

  Since the GMPLS protocol is operating between the OXCs 3 and 4, the protection line (3) can be selected according to the line status and the operator policy. If the protection line (3) is selected according to the line status, the traffic in the network is concentrated, it is possible to avoid the path where congestion occurs, and to distribute the traffic to the path with low utilization rate. Can be used effectively.

  After the switching at the OXCs 3 and 4 occurs, the L2SWs 1 and 2 again set the line (2) and the recovery line as LAG (FIG. 5). Thus, the failure recovery is completed, and service is performed by setting the recovery line using the line (2) and the protection line (3) to LAG.

  FIG. 6 is a functional block diagram showing an embodiment of the L2SW (1 or 2) and OXC (3 or 4) that performs the failure recovery operation. The L2SW includes a failure detection unit 6-1 that detects a line failure, and a LAG switching unit 6-2 that performs LAG setting and cancellation. In addition, the OXC includes a failure detection unit 6-3 that detects a failure in a line, a route calculation unit 6-4 that performs route calculation to obtain another optimum line when a line failure is detected, and a route calculation. A line switching unit 6-5 for switching lines according to the result is provided. These functional units of L2SW and OXC can be realized by software.

  FIG. 7 is a flowchart showing the failure recovery operation in FIG. 6. The flow on the left side of FIG. 6 shows the operation for L2SW, and the flow on the right side shows the operation for OXC.

  As shown in FIG. 2, when a failure occurs in the line (1) (S71), the failure detection units 6-1 and 6-3 of the L2SW and OXC detect this failure (S72 and S75). This failure detection is performed at approximately the same time by L2SW and OXC. When a failure is detected, the LAG switching unit 6-2 of the L2SW immediately switches the traffic on the line (1) to the line (2) for which the LAG is set, and rescues it (S73). Since this switching is performed in about 50 ms, the traffic loss is slight during that time.

  Further, the route calculation unit 6-4 of the OXC performs route calculation when a failure of the line (1) is detected (S76). In this route calculation, an optimum line for performing failure recovery by the OXC restoration function is obtained. Since the GMPLS protocol is operating between the OXCs, it is possible to explicitly select a route according to the line status and the operator policy when the route calculation unit 6-4 calculates the route. In the above example, the protection line (3) is obtained by this route calculation.

  The OXC line switching unit 6-5 switches the line (1) in which the failure is detected to a restoration line using the protection line (3) obtained by the path calculation unit 6-4 (S77). Since it is necessary to perform route calculation by the route calculation unit 6-4, it takes about 500 ms from the time a failure is detected to this switching. Therefore, (LAG switching time in S73) <(OXC switching time in S77).

  After the line is switched by OXC and the line is restored, the LAG switching unit 6-2 of the L2SW performs line switching back by LAG. That is, the recovery line using the line (2) and the protection line (3) is again set to LAG (S74).

  The fault is recovered according to the above flow (S78). According to this, when a failure occurs, traffic is first relieved by the LAG technique. Next, the line is restored by line switching by OXC, and then traffic that has been avoided by LAG flows to the restored line. When a failure occurs, first, the LAG technique is used for high-speed relief, so that traffic loss can be minimized. Further, since there is no need to maintain a protection line between the L2SWs, the L2SW interface can be used effectively.

  In the above embodiment, it is assumed that (LAG switching time) <(OXC switching time) (for example, LAG switching time and OXC switching time are about 50 ms and 500 ms, respectively), and the line by LAG after the line switching by OXC Switching back is performed to obtain a stable operation. However, it is expected that the OXC switching time will be shortened due to future technological advancement, and if the LAG switching time and the OXC switching time are very close, the failure recovery operation may become unstable. Such instability of the failure recovery operation can be prevented by the embodiment described below.

  FIG. 8 is a functional block diagram showing L2SW (1 or 2) and OXC (3 or 4) in this embodiment, and FIG. 9 is a flowchart showing the failure recovery operation. 8 and 9, the same or equivalent parts as those in FIGS. 6 and 7 are denoted by reference numerals.

  FIG. 8 differs from FIG. 6 in that the OXC includes a timer control unit 6-6. FIG. 9 differs from FIG. 7 in that a flow showing the operation for OXC includes a step (S79) of providing a switching protection time by the timer control unit 6-6.

  If the LAG switching time and the OXC switching time are very close to each other as it is, the timer control unit 6-6 may ensure that (LAG switching time) <(OXC switching time). Delay switching time. That is, the timer control unit 6-6 sets the switching protection time in S79, and even if the route calculation by the route calculation unit 6-4 is finished, the line switching by OXC is not performed at that time, and after the switching protection time has elapsed. Make sure that the line is switched. Note that the switching protection time set by the timer control unit 6-6 can be changed so that the minimum necessary switching protection time can be set even if the time required for route calculation changes due to software version upgrade or the like. It is preferable to do so.

  Next, a loop avoidance measure in the present invention will be described. In the case of a network configuration that does not use STP or RSTP, traffic congestion may occur. Therefore, congestion is avoided by allocating a VLAN in units of a plurality of lines set with LAG in L2SW.

  FIG. 10 is an explanatory diagram of loop avoidance by VLAN setting. In this example, the line (1) and the line (2) are set to LAG, and the line (4) and the line (5) are set to LAG. In this case, for example, the VLAN 100 is allocated to the line (1) and the line (2) for which LAG is set, and the VLAN 200 is allocated to the line (4) and the line (5).

  Since the lines (1) and (2) and the lines (4) and (5) are grouped separately by VLAN, no loop is formed. In addition, even if a failure occurs in one of the grouped lines, for example, the line (1) or (2), the traffic can be relieved by the line (2) or (1) configured with LAG. .

It is a block diagram which shows the basic composition of the L2 network to which this invention is applied. It is explanatory drawing of failure generation | occurrence | production and failure detection in this invention. It is explanatory drawing of the switching by LAG in this invention. It is explanatory drawing of the line switching by OXC in this invention. It is explanatory drawing of the line | wire recovery by LAG in this invention. It is a functional block diagram which shows embodiment of L2SW and OXC in this invention. It is a flowchart which shows the failure recovery operation | movement in embodiment of FIG. It is a functional block diagram which shows other embodiment of L2SW and OXC in this invention It is a flowchart which shows the failure recovery operation in embodiment of FIG. It is explanatory drawing of the loop avoidance by the setting of VLAN in this invention.

Explanation of symbols

1, 2 ... L2 switch (L2SW), 3, 4 ... Optical cross-connect device (OXC), 6-1, 6-3 ... Fault detection unit, 6-2 ... LAG switching unit, 6-4: Route calculation unit, 6-5: Line switching unit, 6-6: Timer control unit

Claims (4)

An optical cross-connect device is deployed between the L2 switches,
When the L2 switch detects a line failure, the L2 switch temporarily transfers the traffic of the line where the failure is detected to another line with LAG setting, and the line is restored by LAG setting after line switching by the optical cross-connect device And
The L2 network high-speed failure recovery method, wherein the optical cross-connect device switches a line to a recovery line using a protection line when a line failure is detected. 2. The L2 network high-speed failure recovery method according to claim 1, wherein the optical cross-connect device includes a timer control unit that delays a time from line detection to line switching. 3. The L2 network high-speed failure recovery method according to claim 1 or 2, wherein VLANs are allocated in units of a plurality of lines that are LAG-configured by the L2 switch. 4. The L2 network fast failure recovery method according to claim 1, wherein the protection line is selected in accordance with a line condition or an operator policy.

Images