WO2008028376A1

WO2008028376A1 - A method for improving the fair efficiency in the resilient packet ring and the apparatus thereof

Info

Publication number: WO2008028376A1
Application number: PCT/CN2007/000588
Authority: WO
Inventors: Feng Liu
Original assignee: Zte Corporation
Priority date: 2006-08-30
Filing date: 2007-02-16
Publication date: 2008-03-13
Also published as: CN101136829A; CN101136829B

Abstract

A method for improving the fair efficiency in the resilient packet ring and the apparatus thereof include: a queue management circuit, for storing addresses according to the precedence order queue of packets, and storing the packet destination station addresses accordingly; a calculating circuit (15), for extracting the live time of the use fair rate to obtain the distance from the congestion point to the current station; a scanning circuit (6), for scanning the packet destination station addresses from the first one, to get the packet destination station address; an analysis circuit (7), for analysing the packet destination station address,and determining whether the packet passes through the congestion point based on the distance from the congestion point to the current station. If it does, the scanning circuit (6) is demanded to continue scanning until the packet which doesn't pass through the congestion point is found or the scanning reaches the queue ending; the packet upper ring rate control is performed; when the first packet can't reach the upper ring owing to passing through the congestion point, the packet which doesn't pass through the congestion point is granted to jump the queue to reach the upper ring.

Description

Method and device for improving fairness efficiency in elastic grouping ring

Technical field

The present invention relates to the field of network communication, and in particular to a method and apparatus for improving fairness in an elastic packet ring network. Background technique

The rapid development of Ethernet technology has made data services gradually become the main communication traffic, and data networks have become the development direction of future communication networks. Ethernet adopts the best-effort transmission mechanism, which has good scalability and is suitable for the current bursty data service. However, QoS (Quality of Service) is not guaranteed, and the ability to protect switching is also poor. SDH (Synchronous Digital Hierarchy) devices have a switching time of less than 50ms. There are multiple protection modes and good QoS. However, SDH uses a fixed transmission bandwidth, and the efficiency of transmitting IP data services is not high. Waste, SDH is not the best choice for data service delivery.

Optical Ethernet RPR (Resilient Packet Ring) combines the advantages of Ethernet and SDH. It defines an independent physical layer, the flexible packet ring medium access control layer, which has the advantages of SDH network: Provides a double ring structure. , respectively, the service is transmitted on two rings, with a fast protection mechanism with 50ms protection; at the same time, it has the characteristics of Ethernet: convenient and flexible, plug and play, automatic network discovery, able to adapt to sudden data services, in addition, There are also advantages of spatial multiplexing technology, statistical multiplexing technology and business classification. The spatial multiplexing technology makes full use of the idle bandwidth between any sites on the ring network to improve the bandwidth utilization. The statistical multiplexing technology adjusts the upper ring fair service of the site through the fair algorithm and shares the loop bandwidth. Three types of services: Class A service, Class B service, and Class C service. Among them, Class A service guarantees service bandwidth, delay and jitter, and meets the needs of audio and video services. Class C services do not guarantee loop bandwidth and try to transmit. , to meet the needs of data services.

According to whether the upper ring service is controlled by a fair algorithm, the RPR ring network ring service can be divided into two types of services: fair service (class C service and class B service exceeding the rated rate) and unfair service (class A service and specified rate range). Class B business). The speed guarantee rate of the unfair service, as long as the uplink service does not exceed the guaranteed rate, it is not restricted to the direct ring, but it is never allowed to exceed the warranty. Certificate rate. When the RPR ring network is established, the guaranteed rate of each site is determined. The sum of the guaranteed rates of each site is not greater than the total loop rate, ensuring that the guaranteed rate of each site can be achieved. The difference between the total loop rate and the guaranteed rate is the remaining rate. The remaining rate is shared by all sites on the ring. If the guaranteed rate of a site is not used and is idle, it can also be shared by all sites on the ring. The technology of the present invention is only for fair business, and does not consider non-fair services. The data flows described later are all fair services.

In order to ensure that all stations can enjoy the upper ring rate that can be shared in the loop, the fair frame is defined in the RPR technical standard. It is recommended to use the fair algorithm to calculate the loop traffic and control the upper ring rate of each site to ensure the fairness of each site. The service can distribute the remaining bandwidth on the loop fairly, without the site location and service traffic, causing some sites to occupy more loop bandwidth, and some sites occupy less loop bandwidth, and finally achieve fair distribution of all sites. Loop bandwidth. in

In the RPR standard, the length of the fair period is defined as 100 μ _δ . In a fair period, each station counts the traffic of packets on the ring, and receives, calculates, uses, and sends four fair rate values to achieve fair rate control of the entire loop, ensuring fair access to each site. Loop bandwidth. The four fair rate values are:

Receive Fair Rate: Receives the fair rate sent from the downstream site, and controls the local ring rate of the site to not exceed the fair rate of reception.

Suggested fair rate at this point: The station calculates the recommended fair rate of the site through a set of fair calculation formulas based on the loop traffic, idle bandwidth capacity, and the upper ring rate of the site within a fair period. . In the fair algorithm control, when the recommended rate calculated by one station is smaller than the recommended rate of other stations, other stations on the loop select the minimum rate obtained by the fair algorithm to control, and the station is called a blocking point. The meaning of the blocking point is: The data traffic passing through this site (referred to as this point) is too large, and congestion occurs. As a result, the ring packets on the site are restricted. At this time, the recommended rate calculated by the site is the smallest. This point uses the fair rate: the fair rate used at this point is equal to the smaller one of the fair rate of the site and the fair rate recommended by the site. That is, the fair rate actually used by the site cannot be greater than the fair rate of the receiver, nor can it be greater than this. The recommended fair rate of the site can only be the smaller of the two. Sending a fair rate: By counting the traffic on the local ring and the traffic on the upper ring, you can determine whether the upstream site has any relationship with the downstream traffic. If the upstream site has nothing to do with downstream congestion, upstream If the incoming data packet has been looped and no longer continues to be transmitted along the loop, then the site sends its own proposed fair rate of the proposed point. If the upstream site and the downstream congestion are related, the point is sent using the fair rate. The ring rate on the site cannot be greater than this fair rate.

The site uses the fair rate to control the upper ring rate of the fair service, ensuring that the actual ring rate of the site is not greater than the upper ring rate of the blocking point (this site is related to the blocking point), and avoids occupying more loop bandwidth.

In actual work, each site receives, calculates, uses, and transmits four fair values at the same time, and operates once in each fair period. These operations are synchronized through all sites to achieve a balance of fair services on the ring. Because of the double-ring structure, each site has two sets, which control two loops. There is no relationship between the two loops working independently in the work. In the scenario, the outer ring is introduced as an example. The specific operation process of the site fairness algorithm is as follows:

In Figure 1, the S2 site is taken as an example: the ring packet on the S2 station passes the outer ring (ring 0), and the downstream station S3 sends the downstream recommended rate F3 through the other ring-ring (ring 1). The upper ring rate used to control the S2 site cannot exceed the recommended fair rate F3 given by the downstream site S3. The S2 site calculates the recommended rate F of the site. The fair rate of the S2 site cannot exceed the recommended rate of the site. Nor can it exceed the downstream recommended fair rate F3, whichever is smaller. The S2 station sends the recommended rate F2 of the local station to the upstream site S1. If the upstream site of the S2 is not related to the downstream congestion, the recommended rate F of the site is directly sent, that is, F2 is equal to F. If the upstream site of the S2 is related to the downstream blocked site. Then, F2 is equal to the minimum value of F and F3, and the upper ring rate of the upstream station S1 is required to not exceed the recommended rate F2.

Figure 2 shows the control process of the fair algorithm: The fair rate of the S5 station received by the S5 station is F6, and the fair rate calculated by itself is F5. Because F5<F6, the S5 station selects the F5 fair rate control on the site. Ring rate. Because the upstream site is associated with downstream congestion, the S5 site will be sent to the S4 site using rate F5. The S4 station receives the fair rate F5 of the S5 site, and the self-calculated recommended rate is F4. Because F4 < F5, the S4 station selects the F4 rate to control the upper ring rate of the site. Because the upstream site is associated with downstream congestion, the S4 site sends F4 to the upstream site S3. The S3 station receives the fair rate F4 of S4 and calculates the fair rate F3 of the local station. Because F3>F4, the S3 station selects F4 to control the upper ring rate of the local station. Because the upstream site is associated with downstream congestion, S3 sends F4 to the upstream site S2. Similarly, S2 also calculates the fair rate F2, However, because F2>F4, the S2 site selects F4 to control the upper loop rate of the site. Because the upstream site is related to downstream congestion, S2 sends F4 to S1; and so on, F1>F4, and the SI site sends F4 to S10. Because the SI site, the S2 site, and the S3 site are all related to the downstream blocking, and F1>F4, F2>F4, F3>F4, the SI site, the S2 site, and the S3 site all select the downstream minimum recommended fair rate F4, that is, the blocking. The recommended rate of the point is used to control the ring rate of the local station and send it to the respective upstream. Therefore, the maximum upper ring rate of the S1 station, the S2 station, and the S3 station is F4. In Figure 2, S4, S5, and S6 are all blocking points, but F4 is the smallest and the blocking is the most serious. Therefore, all upstream related sites of the S4 site use F4 instead of F5 and F6.

If the fair rate used by a site is the same as the fair rate of the blocking point, it indicates that the site has a relationship with the blocking point. This site affects the blocking of the blocking point and blocks the transmission path of the site using the same fair value. area. As shown in Figure 2, the S10 site, the S1 site, the S2 site, the S3 site, and the S4 site all use the same fair rate value and belong to a blocked domain. The following sections discuss the case of the same blocked domain.

The fairness algorithm requires that all sites in the blocked domain control the upper loop rate according to the recommended rate of the blocking point. This ensures that each site can share the loop bandwidth, avoiding that some sites occupy more loop bandwidth, while other sites occupy less loops. Unfairness in bandwidth. However, there is a defect in adopting this control method: When the fairness algorithm performs control, the total amount of all data traffic of the upper ring is controlled. When controlling the bandwidth of the upper ring, the packet traffic that can be controlled without control is also controlled. In Figure 2, because F1>F4, F2>F4, F3>F4, the three sites themselves suggest that the upper ring rate is very large and can be larger than the bandwidth of F4. However, because F4 is small and blocking occurs at the S4 site, the S1-S3 sites can only select F4 to control the ring rate of the site, which avoids the ring rate on the S4 site being too small, resulting in unfairness. When the S1 site has a traffic flow, it is sent from the SI site to the S3 site, as shown in Figure 2. Since the S1 site uses the downstream fair rate F4 to control the upper ring rate bandwidth, the maximum rate of the traffic flow Flow can only be F4. However, in fact, because the blocking occurs at the S4 site, the traffic flow does not pass through the S4 site, but is terminated after the S3 site is looped down. However, the current fair algorithm implementation method is flawed, resulting in a fair rate of using the blocking point F4. To control the traffic flow that does not pass through the blocking point, the traffic on the traffic flow cannot be increased. Summary of the invention

The technical problem to be solved by the present invention is to provide a method and device for improving fairness efficiency in an elastic packet ring, which is used for solving a fair traffic rate that does not pass through a blocking point by using a fair rate of a blocking point, and the service traffic cannot be improved. Shortcomings.

An apparatus for improving fairness and efficiency in an elastic packet ring, comprising:

a queue management circuit for queuing the actual storage address of the data packet in a column of the queue according to the order in which the data packets arrive, and in the other column of the queue, the data is queued according to the queued order of the data packet The destination site address of the packet is queued;

a scanning circuit, connected to the queue management circuit, for scanning a destination site address of each data packet from a first data packet in the queue;

An analysis circuit, connected to the queue management circuit and the scanning circuit, configured to analyze the purpose according to the distance information between the received blocking point and the local station, and the destination site address of the data packet scanned by the scanning circuit Whether the data packet corresponding to the site address passes through the blocking point, and controls the scanning mode of the scanning circuit according to the analysis result;

a proposed fair rate control circuit, configured to connect to the queue management circuit, configured to control an upper ring rate of the data packet according to a recommended fair rate of the site, such that the upper ring rate does not exceed the recommended fair rate, and Allowing a packet ring that does not pass through the blocking point when all the upper ring rates do not exceed the recommended fair rate;

Using a fair rate control circuit, the queue management circuit is configured to control an upper loop rate of the data packet according to a fair rate of use of the site, so that an upper loop rate of the data packet passing through the blocking point does not exceed the use Fair rate; and

a computing circuit, connected to the analysis circuit, the using fair rate control circuit, for extracting a lifetime according to the use fair rate, and obtaining a distance between the blocking point and the site according to the lifetime, and then The distance information is provided to the analysis circuit.

The device for improving fairness in the flexible packet ring further includes a scheduler, and is connected to the queue management circuit, and is configured to select a queue to which the data packet for performing the upper ring operation belongs when the site has multiple queues. ^; Fair rate the recommended control packet comprises a control circuit without blocking data points; The data packet controlled by the fair rate control circuit includes a data packet passing through the blocking point and a data packet not passing through the blocking point.

The apparatus for improving fairness efficiency in the resilient packet ring of the present invention, when the usage fair rate is generated by the local station, the lifetime is 255; when the usage fair rate is by the downstream site of the site The sent time is less than 255.

The distance between the blocking point and the local station is the difference between the survival time and 255; when the difference is 0, the blocking point is the local station, when the difference is N Then, the blocking point is the Nth station after the local station, where N is a natural number greater than or equal to 1.

A method for improving fairness efficiency in an elastic packet ring by using the device of the present invention, characterized in that it comprises:

Step 1: The queue management circuit queues the actual storage address of the data packet in one column of the queue, and queues the destination site address of the data packet according to the queue order of the data packet in another column of the queue. ;

Step 2: The calculation circuit extracts a lifetime according to the use fair rate, and obtains a distance between the blocking point and the site according to the survival time;

Step 3: The analyzing circuit analyzes, according to the distance, the destination site address of the data packet scanned by the scanning circuit, whether the data packet corresponding to the destination site address passes through a blocking point, and controls the scanning circuit according to the analysis result. Scanning method; and

Step 4: The recommended fair rate control circuit controls an upper ring rate of the data packet, so that all the upper ring rates do not exceed the recommended fair rate, and when all the upper ring rates do not exceed the recommended fair rate Allowing a packet to be looped through the blocking point; controlling the upper ring rate of the data packet by using the fair rate control circuit such that the upper ring rate of the packet passing through the blocking point does not exceed the usage fair rate.

In the third step, when the data packet corresponding to the destination site address passes through the blocking point, the analyzing circuit controls the scanning circuit to continue scanning the queue until the data packet that does not pass through the blocking point is scanned or scanned to the tail of the queue.

In the step 4, when the first data packet in the queue passes a blocking point and the upper ring rate of the data packet currently passing the blocking point is greater than the usage fair rate, the first data is sent. The packet entry rate drops waiting for the upper ring state.

In the fourth step, when the upper ring rate of all the upper ring data packets is less than the recommended fair rate of the local station and the first data packet is in the waiting upper ring state, the data packet is not allowed to pass through the blocking point. Sheung Wan.

In the step 4, after the data packet that has not passed through the blocking point is queued, the queue management circuit deletes the data packet from the queue to which the data packet belongs, and the scanning circuit continues to scan the data packet. data pack.

The present invention solves the disadvantage of using a fair rate of blocking points to control fair traffic that does not pass through the blocking point, resulting in an inability to improve traffic.

Drawing fan

Figure 1 is a schematic diagram of the fair algorithm control;

2 is a schematic diagram of a control process of a fair algorithm;

Figure 3 is a schematic diagram of a fair frame structure; '

4 is a schematic diagram of an apparatus for managing a ring-up data packet at each station;

Figure 5 is a schematic view of the apparatus of the present invention;

Figure 6 is a flow chart of the method of the present invention. Preferred embodiment of the invention

Figure 3 shows a schematic diagram of the fair frame structure. In the application of the fairness algorithm, each site uses the fair rate of the blocking point to control all the upper ring data packets in the blocking domain, so that the data packets in the ring data packets of the local station that are not blocked are also controlled by the fair rate. Upper loop rate and transfer efficiency. In practical applications, the fair rate of the blocking point is transmitted in the form of a fair frame, which is transmitted from the downstream site to the upstream site. When the recommended rate of a site is greater than the fair rate of the blocking point, the site uses the fair rate of the blocking point (the site finds no free bandwidth after the statistical loop condition), and forwards the fair rate of the blocking point to the upstream. transfer. The fairness structure of the fair rate is shown in Figure 3. The frame contains: time to live (LiveToLive), ring (road) control (baseRingControl), source (site) address (sourceMacAddress), fair-head check/fair control header (faimessControlHeader), fair control value (faimessControlValue) and frame check sequence (frameCheckSequence).

In a fair frame, there are three key parts: time to live, source address, and fair (control) value. Time to live: Gives the recommended fair rate to pass the number of sites on the loop. When the fair rate is generated, the value is 255. If the fair rate of delivery remains the same, the value is decremented by one each time a site passes.

Source Address: Gives the site address that generated the fair rate, which is the site address of the blocking point. Fair (Control) value: This part conveys the recommended fair value of the downstream site, that is, the fair rate of the blocking point, which is used to control the upper ring data flow of the upstream site. .

When a new fair rate is generated and a fair rate is passed upstream, the initial value of the lifetime is

255. If the upstream site wants to continue uploading this fair value, it will keep the fair value, the source address unchanged, and the lifetime reduced by 1. When the other sites receive the fair value, they can know the source site address of the fair value. The size of the time-to-live can be used to know the distance from the source site, that is, the distance from the blocking point. In Figure 2, the S4 station calculates the fair rate as F4, F4<F5, so S4 sends F4 to S3, and the sent fair frame contains the site information of F4 and S4, and the lifetime is 255. After receiving the fair frame sent by the S4 station, the S3 station decrements the lifetime 255 by 1 and becomes 254. Because F3>F4, the S3 station continues to send F4 to the S2 site. The fair frame rate is still F4, the source site address is S4, and the lifetime is 254 after the subtraction operation. Similarly, after the S2 station receives the subtraction operation, the fair rate sent to the S1 site is still F4, and the survival time becomes 253; after the S1 station receives the subtraction operation, the fair rate sent to the S10 site is still F4, and the survival time becomes Into 252.

Through the time-to-live, you can know the address of the site that generates the fair value from the site, that is, the distance from the site to the site. In the case of knowing the destination site of the ring packet on the site, if the upper ring data packet does not pass through the blocking point, as shown in FIG. 2, the ring data flow F w at the site S1 does not pass through the blocking point S4, then the data flow can be It is not controlled by F4, but only controlled by F1 calculated by this site. Because F1>F4, the upper loop bandwidth of the data flow Flow will increase, which improves the efficiency of packet transmission. The specific implementation is as follows: At each site, the queue management technology is used to classify all the upper ring data packets and queue them according to the categories. The data packets in the same queue are queued in the upper ring in order. The data packets between different queues have no order relationship. The upper ring sequence between the queues is controlled by the scheduler. The scheduler determines which of the current queues can be looped, and which queue data packets cannot be looped. The scheduler selects the queue according to its own priority. When a queue is selected, the same queue is internally looped in the order of priority. However, whether it can be on the ring, it is also necessary to judge whether the queue data packet is controlled by the fair algorithm: if it is not controlled by the fair algorithm, it is directly on the ring (such as non-fair business). The invention relates only to fair services, so the queue is controlled by a fair algorithm. When the queue is controlled by the fair algorithm, if the upper loop speed is lower than the fair rate, the ring can be looped. If the upper loop speed is greater than the fair rate, the ring cannot be looped. As shown in Figure 4. '

As shown in FIG. 4, a schematic diagram of a device for managing the upper ring data packet of each station is given. Classify all the upper ring data, queue them in their own queues by class, and wait for the upper ring to send. The actual storage address of the upper ring data packet is stored in each queue, and is stored in order. The head pointer indicates the data packet storage address with the longest queue time, and the tail pointer indicates the data packet storage address with the shortest queue time. The queue can store the specific 'content of the data packet, or only the actual storage address of the data packet. The specific content of the data packet may be stored in other places, such as a memory module. The present invention introduces the actual storage address of the data packet stored in the queue as an example. The actual storage address of the data packet is only stored in the queue, and the actual storage address represents the content of the data packet, and the actual storage address of the data packet is performed in the queue. Queuing, which means that packets are queued in the queue. When you need to find the contents of the data packet, first find the actual storage address of the data packet in the queue, and indirectly search for the data packet content through the actual storage address. The fairness algorithm is used to control whether the upper ring data packet is greater than the fair rate. If 杲 is greater than, the data packet is forbidden to the upper ring. The scheduler is used to select which queue data is currently available for ringing. The scheduler is used in multiple queue devices, and the scheduler is used to indicate that the queue can perform the ring operation. The fair algorithm control and the scheduler work together to determine whether the data packets in the queue are passed on the ring. If the scheduler dispatches to the queue (the queue can perform the upper loop operation) and can be looped under the control of the fair algorithm, then one data packet can be looped, followed by the second, third, ... data packets. Even if the scheduler dispatches to this queue, if the traffic is too large under the control of the fair algorithm and the upper ring is not allowed, the first data packet cannot be looped. If the first packet cannot be looped, all subsequent packets cannot be looped. In the queue management, the queue is controlled according to the fairness algorithm. The upper ring is sent in the order, the previous data packet is not sent to the upper ring, and the subsequent data packets cannot be sent to the upper ring. Under the control of the fairness algorithm, because the upper ring is sent in order, regardless of whether the data packet passes through the blocking point, the data packet that does not pass through the blocking point later may not be looped due to the previous data packet. Packet delivery efficiency.

As shown in FIG. 5, it is a schematic diagram of the apparatus of the present invention. In the present invention, in addition to the storage address of the stored data packet in each queue, the transmission destination station address and other attached circuits of one column of data packets are added, so that the queue management structure becomes the device structure shown in FIG.

The apparatus structure includes: a queue management circuit 51, a scanning circuit 52, an analysis circuit 53, a proposed fair rate control circuit 54, a fair rate control circuit 55, a calculation circuit 56, and a scheduler 57.

The queue management circuit 51 queues the data packets (actually queued according to the storage address of the data packets) and simultaneously queues the destination site addresses of the data packets.

In the apparatus structure of the present invention, one column in each queue stores the storage address of the queued data packet, and the other column corresponds to the destination site address of the queued data packet in the order in which the data packets are queued. The store address information is used to query the storage location of the data packet. The destination site address indicates to which site the data packet is sent to analyze whether the data packet has passed the blocking point.

The scanning circuit 52 is configured to scan the destination site address of each data packet; according to the requirement of the analysis circuit 53, start scanning from the first data packet until scanning the first data packet that does not pass through the blocking point, or scans the team tail.

The analyzing circuit 53 is configured to analyze the destination site address of the data packet scanned by the scanning circuit 52, and analyze the data packet corresponding to the destination site address of the data packet according to the distance information between the blocking point provided by the calculating circuit 56 and the local station. Whether it passes through the blocking point. If the blocking point has elapsed, the scanning circuit 52 is required to continue scanning backward from this point until the first packet that does not pass through the blocking point is scanned, or scanned to the end of the queue.

The fair rate control circuit 54 is configured to control the upper ring rate of the data packet according to the recommended fair rate of the site, and the upper ring rate of all the upper ring data packets cannot be greater than the recommended fair rate; if all current upper ring rates are not Greater than the recommended fair rate of this site, it is allowed to pass without blocking The data packet on the slug is on the ring (the packet is not controlled by the fair rate control circuit). In the present invention, the control is suitable for data packets that pass through and do not pass through the blocking point.

Using the fair rate control circuit 55, for controlling the upper loop rate of the data packet according to the usage fair rate of the site, requiring that the data packet upper loop rate passing through the blocking point cannot be greater than the usage fair rate, in the present invention, Controls packets that are appropriate for all blocked points.

The calculating circuit 56 is configured to obtain a survival time according to the use fair rate, calculate a distance of the blocking point from the site according to the survival time, and provide the distance information to the analyzing circuit 53, and the analyzing circuit 53 analyzes whether the data packet passes through the blocking point.

If the fair rate is generated by the site, the lifetime is 255; if it is generated by other sites, it is less than 255.

If the lifetime of the fair frame applied to the site is 255, the blocking point is the site; if the lifetime is 'j, 255, indicating that the blocking point is at a certain point downstream, the distance of the blocking point from the site is equal to the The difference between the survival time and 255. In this example, the distance extracted by the S1 site is 3 (the lifetime of the S4 site is 255, and it is reduced to 252 when it is delivered to the S1 site), indicating that the distance between the blocking point and the S1 site is 3.

The calculation circuit 56 calculates the difference between the survival time and 255 to obtain the distance of the blocking point from the site, and the analyzing circuit 53 analyzes the distance information.

The scheduler 57 is used to select which queue of packets can be looped. The scheduler is only valid in the case of multiple queues. If there is only one queue at this site, the scheduler is omitted.

Taking the site S1 as an example, in the specific work, the head pointer indicates the first data packet to be sent, and the data packet is the first queued data packet, and needs to be sent first. If the queue is controlled by the fair algorithm rate, the fair rate used by the station is F4, the first packet passes through the blocking point, and the current upper loop traffic is greater than the fair rate F4, then control is required to prohibit the packet from ringing up to the upper ring. The flow is reduced before it can be looped. During the prohibition of the first packet ring, the scanning circuit 52 scans the destination site address of each packet from the head of the queue to analyze whether the destination site address passes through the blocking point. When the second data packet is scanned, it is found that the destination address of the second data packet does not pass through the blocking point S4, but is aborted before the blocking point, as in this example, the data flow Flow is terminated at the S3 station, so that The data packet can be controlled without using the fair rate F4, and only needs to be recommended by this point. Rate Fl control can be. In this case, the first queued data is controlled by the fair rate F4, and the first packet is not allowed to ring, but after that, the second packet does not pass through the blocking point and can be controlled without using the fair rate F4. Fair rate F1 control. If the fair rate F1 is allowed, the second packet can be inserted into the ring, and the idle time of the upper ring can be fully utilized to queue the second packet. Because F1>F4, the data packet can be looped up with F1 traffic, and the upper loop speed increases.

Initially, scanning circuit 52 scans from the head of the team until the first packet that does not pass through the blocking point is scanned, or is not found at the end of the scan and stays there. If the first queued packet in the queue cannot be looped because it uses fair rate control, the first packet that does not pass through the blocking point is allowed to ring on the ring in idle time, and the packet that does not pass through the blocking point is queued. Not subject to the order of the queue. The packet on the queue is not controlled by the fair rate used at this point (the fair rate is used in the S1 site in the figure is F4), but only the recommended fair rate control calculated by the site (the S1 site recommended fair rate is Fl in the figure). Because the recommended rate value calculated by the site uses the fair rate value, the rate of the upper ring data packet increases, and the loop transmission efficiency is improved.

After the packet on the upper ring of the queue is looped, the packet needs to be deleted from the queue to avoid looping again, causing two loop errors. The delete operation is implemented by the queue management function of the queue management circuit 51. After the delete operation, the scan circuit 52 continues the scan of the subsequent packets until the next packet that does not pass through the blocking point is scanned. Since the network packet traffic is dynamically changing, the blocking point shifts as the traffic on the ring network changes. The original blocking point may become a non-blocking point, and the original non-blocking point can become a new blocking point. When the blocking point transitions, the length of the blocking point and the site will change. If the distance from the blocking point to the site increases, the packets that have been scanned from the beginning of the queue may change the packets that have passed through the blocking point to not pass through the blocking point. These new packets that do not pass through the blocking point. The upper ring can be re-plugged; if the distance from the blocking point is reduced, the packet that has passed through the blocking point still passes through the blocking point, and the scanning can be continued in the previous manner regardless of the already scanned data packet. Just use the new blocking point distance information. Therefore, in practical applications, it is only necessary to consider the case where the blocking point and the site are increased in distance. The analysis circuit 53 will detect this when the blocking point distance becomes larger. Upon detecting an increase in the blocking point distance, the analysis circuit 53 requests the scanning circuit 52 to rescan from the head of the team to redetermine the position of the packet that does not pass through the blocking point. In operation, if the scanning circuit 52 points to the first data packet, it means that the first data packet in the queue does not pass through the blocking point, and can be directly uplinked, and the calculation is performed according to the recommended rate of the point, instead of using the fair rate upper ring. . In this way, it can be guaranteed that the data packet does not occupy the fair rate of use, but only takes up the recommended fair rate. In this case, after the first packet is queued, the head pointer points to the next packet, and the scanning circuit 52 scans from the next packet, scanning out the new first one without blocking. Point of the packet.

As shown in FIG. 6, it is a flowchart of the method of the present invention. The method includes the following steps: Step 61: In the data packet queuing queue, the queue management circuit queues the address according to the sequence of the data packets, and correspondingly stores the destination site address of the data packet;

The sequence refers to sorting according to the order in which the data packets arrive at the queue management circuit. Step 62: When each station uses the fair rate, the computing circuit extracts the lifetime of using the fair rate, and gives the large distance of the blocking point from the site;

Step 63: The scanning circuit starts scanning from the destination site address of the first data packet, and gives the destination site address of the data packet;

Step 64: The analysis circuit analyzes the destination site address of the data packet, and determines whether the data packet passes through the blocking point according to the distance of the blocking point from the local station point. If the 杲 passes through the blocking point, the scanning circuit is required to continue scanning until the data packet that does not pass through the blocking point is scanned, or scanned to the end of the queue;

Step 65: Perform ring-to-ring rate control on the data packet; when: a data packet cannot be looped because it passes through the blocking point, the data packet that does not pass through the blocking point is allowed to be queued.

In the above step 61, a queue queue content is composed of two partial groups: the storage address of the data packet and the destination site address of the data packet, and are queued according to the order of the data packets;

The storage address of the data packet gives the storage location of the data packet. When the data packet is sent, the data packet is requested according to the storage address; the destination site address of the data packet indicates to which station the data packet is sent;

The storage address of the data packet and the destination site address of the data packet (in order) are maintained - the corresponding relationship;

The two parts of the storage address of the data packet and the destination site address of the data packet can be stored separately so as to be operated separately, but the two need to maintain a one-to-one correspondence. In the above step 62, the fair rate used by the site refers to the fair rate of use adopted by the site and controlled by the upper ring rate;

When the fair rate is generated by the site itself, its lifetime is 255; when the fair rate is sent by other downstream stations, its lifetime is less than 255;

The distance from the blocking point to the site is the difference between the time to live and 255. When the difference is 0, it indicates that the blocking point is the local station; when the difference is 1, it indicates that the blocking point is the next adjacent station of the site, and there is only one span from the site; the rest, and so on, That is, when the difference is a natural number N, it means that the blocking point is the Nth station after the site.

In the foregoing step 65, the data packet cannot be looped because it passes through the blocking point, which means that the first data packet passes through the blocking point, and the current ring rate of the data packet passing through the blocking point is greater than the fair rate, so that the first data packet cannot be used. Ring, you need to wait for the rate to drop before you can go to the ring;

Allowing the ring to not pass through the blocking point packet means that the upper ring rate of all the upper ring data packets is less than the recommended fair rate of the site itself (that is, the recommended fair rate used at this point), and the first data packet needs to wait. Ring, which allows packets that do not pass through the blocking point to be queued to the upper ring;

After the packet that has not passed through the blocking point is queued, the packet is deleted from the queue, and the scanning circuit continues to scan the subsequent data packet to prepare for the next queue.

The present invention applies a fairness algorithm for statistical multiplexing and spatial multiplexing, and fully utilizes the remaining bandwidth on the loop to achieve fair access to these loop bandwidths at all sites on the loop. The present invention determines the location of the blocked site by analyzing the information provided by the fairness algorithm while maintaining the fairness algorithm unchanged. Analyze the destination site address of each packet in the queue management to determine if the packet has passed the blocking point. In the case of the control of the fair algorithm, if the first data packet in the queue cannot be sent out, the data packet in the queue that does not pass through the blocking point is allowed to be inserted into the ring, thereby improving the transmission efficiency of the fair service of the upper ring. The present invention solves the disadvantage of using a fair rate of blocking points to control fair traffic that does not pass through the blocking point, resulting in an inability to improve traffic. .

The invention may, of course, be embodied in a variety of other embodiments without departing from the spirit and scope of the invention. Changes and modifications are intended to be included within the scope of the appended claims. Industrial and practical 4 students

The present invention applies a fairness algorithm for statistical multiplexing and spatial multiplexing, and fully utilizes the remaining bandwidth on the loop to achieve fair access to these loop bandwidths at all sites on the loop. The present invention determines the location of the blocked site by analyzing the information provided by the fairness algorithm while maintaining the fairness algorithm unchanged. Analyze the destination site address of each packet in the queue management to determine if the packet has passed the blocking point. In the case of the control of the fair algorithm, if the first data packet in the queue cannot be sent out, the data packet in the queue that does not pass through the blocking point is allowed to be inserted into the ring, thereby improving the transmission efficiency of the fair service of the upper ring.

Claims

Claim

A device for improving fairness and efficiency in an elastic packet ring, comprising: a queue management circuit, configured to queue the actual storage address of the data packet according to the order of arrival of the data packets in a queue of the queue, and Queue the destination site address of the data packet in another queue of the queue according to the queue order of the data packets;

a scanning circuit, connected to the queue management circuit, for scanning the first packet from the first data packet in the queue: a destination site address of the packet;

An analysis circuit, connected to the queue management circuit and the scanning circuit, configured to analyze the distance information according to the received blocking point and the local station, and the destination site address of the data packet scanned by the scanning circuit Whether the data packet corresponding to the site address passes through the blocking point, and controls the scanning mode of the scanning circuit according to the analysis result;

a proposed fair rate control circuit, configured to connect to the queue management circuit, to control an upper loop rate of the packet according to a recommended fair rate of the site, so that all the upper loop rates do not exceed the recommended fair rate, And allowing the data packet ring that does not pass through the blocking point when all the upper ring rates do not exceed the recommended fair rate;

2. The apparatus for improving fairness efficiency in an elastic packet ring according to claim 1, further comprising a scheduler, connected to the queue management circuit, configured to select from when the site has multiple queues. The queue to which the packet of the upper ring operation belongs.

3. The apparatus for improving fairness efficiency in an elastic packet ring according to claim 1 or 2, wherein the data packet controlled by the recommended fair rate control circuit includes data that does not pass through a blocking point; and the use is fair. The data packets controlled by the rate control circuit include data packets passing through the blocking point and data packets not passing through the blocking point.

4. The apparatus for improving fairness efficiency in an elastic packet ring according to claim 1 or 1, wherein when said use fair rate is generated by the site, said lifetime is 255; when said using The fair rate is sent by the downstream site of the site, and the lifetime is less than 255.

5. The apparatus for improving fairness efficiency in an elastic packet ring according to claim 4, wherein a distance between the blocking point and the local station is a difference between the survival time and 255; When the value is 0, the blocking point is the local station. When the difference is N, the blocking point is the Nth station after the local station, where N is a natural number greater than or equal to 1.

6. A method for improving fairness efficiency in an elastic packet ring by using the apparatus of claim 1. The method comprises:

Step 4: The recommended fair rate control circuit controls the upper ring rate of the data packet, so that all the upper ring rates do not exceed the recommended fair rate, and when all the upper ring rates do not exceed the recommended fair rate Allowing a packet to be looped through the blocking point; controlling the upper ring rate of the data packet by using the fair rate control circuit such that the upper ring rate of the packet passing through the blocking point does not exceed the usage fair rate. '

The method for improving fairness efficiency in the resilient packet ring according to claim 6, wherein in the step (3), when the data packet corresponding to the destination site address passes through the blocking point, the analyzing circuit control station The scanning circuit continues to scan the queue until a packet that does not pass through the blocking point is scanned or scanned to the end of the queue.

The method for improving fairness efficiency in an elastic packet ring according to claim 6 or 7, wherein in step 4, when the first data packet in the queue passes a blocking point and When the upper ring rate of the data packet passing through the blocking point is greater than the usage fair rate, the first data packet entering rate decreases and waits for the upper ring state.

The method for improving fairness efficiency in the resilient packet ring according to claim 8, wherein in the fourth step, when the rate of the upper ring of all the upper ring data packets is smaller than the recommended fair rate of the site, When the first data packet is waiting for the upper ring state, the data packet that does not pass through the blocking point is allowed to be queued.

The method for improving fairness in the resilient packet ring according to claim 9, wherein in the step 4, when the data packet that does not pass through the blocking point is queued, the queue management circuit The data packet is deleted from the queue to which the data packet belongs, and the scanning circuit continues to scan the data packet after the data packet.