WO2007051374A1 - A method for guaranteeing classification of service of the packet traffic and the method of rate restriction - Google Patents
A method for guaranteeing classification of service of the packet traffic and the method of rate restriction Download PDFInfo
- Publication number
- WO2007051374A1 WO2007051374A1 PCT/CN2006/001012 CN2006001012W WO2007051374A1 WO 2007051374 A1 WO2007051374 A1 WO 2007051374A1 CN 2006001012 W CN2006001012 W CN 2006001012W WO 2007051374 A1 WO2007051374 A1 WO 2007051374A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- packet
- class
- service
- rate
- ring
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000005538 encapsulation Methods 0.000 claims abstract description 39
- 238000012544 monitoring process Methods 0.000 claims abstract description 13
- 238000009789 rate limiting process Methods 0.000 claims description 9
- 238000005516 engineering process Methods 0.000 description 17
- 230000005540 biological transmission Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 3
- RGNPBRKPHBKNKX-UHFFFAOYSA-N hexaflumuron Chemical compound C1=C(Cl)C(OC(F)(F)C(F)F)=C(Cl)C=C1NC(=O)NC(=O)C1=C(F)C=CC=C1F RGNPBRKPHBKNKX-UHFFFAOYSA-N 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011144 upstream manufacturing Methods 0.000 description 2
- 101100234002 Drosophila melanogaster Shal gene Proteins 0.000 description 1
- 235000015076 Shorea robusta Nutrition 0.000 description 1
- 244000166071 Shorea robusta Species 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000001404 mediated effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/25—Flow control; Congestion control with rate being modified by the source upon detecting a change of network conditions
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
- H04L12/42—Loop networks
- H04L12/437—Ring fault isolation or reconfiguration
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/11—Identifying congestion
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/215—Flow control; Congestion control using token-bucket
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
- H04L47/2408—Traffic characterised by specific attributes, e.g. priority or QoS for supporting different services, e.g. a differentiated services [DiffServ] type of service
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10—TECHNICAL SUBJECTS COVERED BY FORMER USPC
- Y10S—TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10S370/00—Multiplex communications
- Y10S370/908—Local area network
- Y10S370/909—Token ring
Definitions
- the invention relates to a data transmission technology for a data packet service, in particular to a method for guaranteeing a monthly service level of a data packet service on a Resilient Packet Ring (RPR) and a rate restriction method based on a token bucket.
- RPR Resilient Packet Ring
- RPR ring network is a transmission technology for data packet services. It combines unique carrier-class service features with Ethernet's high-bandwidth distribution, flexibility and scalability for data services to provide operators with data optimization. Bandwidth management and cost-effective multi-priority service delivery solutions.
- FIG 1 shows the structure of an existing RPR ring network and the node structure on the RPR ring network.
- the RPR ring network is a reverse-oriented double-ring structure, which is divided into inner ring (also known as ringletl) and outer ring (also known as ringletO), which can support up to 255 nodes, as shown in Figure 1.
- inner ring also known as ringletl
- outer ring also known as ringletO
- S0 ⁇ S254 the inner and outer rings on the RPR ring network can complete the transmission and reception of data frames.
- the node on the RPR ring network is composed of a user-side physical medium access control client (MAC client), a physical layer entity, and a medium access control (MAC) entity, wherein the MAC entity further includes a MAC control entity and two respectively
- the outer ring data channel entity and the outer ring data channel entity associated with the outer ring may further be divided into a west physical layer entity and an east physical layer entity.
- the outer ring receives data from the west to the physical layer entity, and sends data through the east physical layer entity.
- the inner ring receives data from the east to the physical layer entity, and sends data through the west physical layer entity.
- the MAC client can pass the data of the inner and outer rings.
- the channel sends and receives data.
- the RPR MAC entity needs to provide three types of service data services for the user-side entity: class A, class B, and class C.
- the characteristics of the three service level services are as follows:
- Class A service is a service with bandwidth and jitter.
- the class A service can be further subdivided into class AO and class A1 services within the RPR MAC entity.
- the bandwidth of the class AO service is pre-allocated and is not allowed to be occupied by other priorities and services of other nodes. It is mainly used to carry real-time services with strict requirements on delay and jitter.
- the class A1 service is characterized by its bandwidth and jitter, but its bandwidth is not reserved. In the case that the service has no transmission demand, the bandwidth is It can be reclaimed by other priorities and other nodes' services, thus improving ring network bandwidth utilization, but its access delay is greater than class A0.
- the class B service can be further subdivided into class BO and class B1 services within the RPR MAC entity.
- the class B0 service is a service with guaranteed bandwidth, jitter and delay guarantee, and is also called within the access commitment rate.
- Class B—CIR service which can be reclaimed by low-priority services without a transmission requirement;
- class B1 service is a class B service whose rate exceeds the committed rate, and is a class B-EIR service, which is called class B-EIR service. Its bandwidth, jitter and latency are not guaranteed.
- Class C service is a service that tries to forward, and its bandwidth, jitter and delay are not guaranteed.
- the RPR ring network needs to rate all services of the upper ring and different service levels. This rate limitation is to achieve various service levels in the ring network as expected. The necessary prerequisite for allocating bandwidth to share network resources.
- the specific method of the rate limiting is to accurately obtain the link bandwidth of the physical layer, allocate the physical layer bandwidth to different nodes and corresponding priorities according to the pre-planning, and ensure the bandwidth of all service level services requiring bandwidth guarantee. The sum of the configurations is not greater than the bandwidth of the physical link. In this way, the bandwidth of the corresponding various services is guaranteed. Under the premise that the bandwidth of various services is guaranteed, the jitter and delay can be determined by the scheduling mechanism of the RPR MAC entity. N2006/001012 is guaranteed.
- RPR is a data link layer technology
- the protocol standard requires that its physical layer implementation can be carried to different physical layer entities.
- digital synchronization series (SDH/Sonet) physical layer technology and gigabit can be selected.
- the RPR packet needs to be specially encapsulated or added with a certain overhead before the data transmission can be realized.
- the transmitted data packet such as RPR message
- the data packet to be transmitted for example, an RPR packet
- the data packet to be transmitted needs to be further added with a preamble (Preamble) and a frame start delimiter (SFD, Start Frame Delimiter) before being transmitted.
- Preamble a preamble
- SFD Start Frame Delimiter
- the actual bandwidth obtained by the RPR MAC entity is less than its nominal bandwidth. This type of bandwidth loss is likely to result in a forwarding service with a lower service level.
- the bandwidth of the service with a higher service level on the upper ring of the node is preempted.
- the class forwarded by one node may preempt the class AO service on the ring of the node, or the class C service forwarded by one node may also preempt the class AO service on the ring of the node, so that the true and expected service level guarantee cannot be achieved.
- the present invention provides a method for guaranteeing a service level of a data packet service, which avoids a situation in which a low service level forwarding service preempts a local high service level upper ring service when a link is congested. , to achieve a true, expected level of business service.
- Another object of the present invention is to provide a rate limiting method that can accurately control the statistical rate of data packet traffic within a nominal bandwidth.
- each node on the resilient packet ring performs the following steps:
- rate limiting is performed according to the physical packet length of the flexible packet ring of each service level service
- the rate limit of the physical packet length of the flexible packet ring according to the service level of the service is as follows: In the process of using the token bucket for rate limiting, each time an elastic packet ring message is sent or a flexible one is sent Before the packet ring packet is received, the number of tokens in the current token bucket is subtracted from the physical layer encapsulation overhead of the elastic packet ring packet.
- the rate limiting using the token bucket in the present invention includes:
- a predetermined number of tokens are added to the token bucket, and the number of tokens in the token bucket is limited to be less than or equal to a preset token high threshold.
- n is less than or equal to 256, and the number of tokens in the current token bucket is subtracted by n;
- the current elastic packet ring message is stopped.
- the rate monitoring of all non-class AO services forwarded by the local node and ringed on the local node is as follows: According to the elastic packet ring physical packet length of the service level of each service class, the ring is forwarded by the local node and is ringed on the local node. All non-performing rate d ass AO traffic statistics, statistics give rates of all non-class AO service;
- each non-class AO service packet is forwarded or The rate of all non-class AO services is added to the physical layer encapsulation overhead of the transmitted packets before each non-class AO service is forwarded.
- the statistic rate of all non-class AO services is added to the physics of the sent packets before each non-class AO service packet sent by the local ring is sent or before each non-class AO service packet sent by the local ring is sent.
- Layer package overhead length The statistic rate of all non-class AO services is added to the physics of the sent packets before each non-class AO service packet sent by the local ring is sent or before each non-class AO service packet sent by the local ring is sent.
- the token bucket performs rate limiting according to the physical packet length of the packet sent by the next logical layer encapsulation process.
- the packet to be sent is currently sent.
- the number of tokens in the current token bucket is subtracted from the length of the overhead used by the next logical layer to encapsulate the packet.
- the rate limiting process of the present invention includes:
- a predetermined number of tokens are added to the token bucket, and the number of tokens in the token bucket is limited to be less than or equal to a preset token high threshold;
- the number of tokens in the current token bucket is subtracted from the number of bytes sent; when the number of tokens in the current token bucket is less than or equal to the preset token low threshold, the packet is stopped. Send the current message.
- the method of the present invention performs rate limiting on the services of the ring on each node according to the RPR physical packet length of the RPR packet and the physical packet length of the RPR packet according to the RPR packet on each node.
- the ring rate or the rate monitoring of all the non-class AO services forwarded by each node can effectively solve the problem that the forwarding service caused by the physical layer encapsulation overhead of the RPR ring network preempts the ring-band service bandwidth of the downstream node, and thus cannot guarantee the reserved bandwidth. defect.
- the technical solution proposed by the present invention is applicable to the random packet length, and does not need to consider the bandwidth loss caused by different packet lengths, and can be allocated according to the physical link bandwidth, so that each node can obtain the expected allocated physical link bandwidth.
- FIG. 1 shows the structure of an existing RPR ring network and the node structure on the RPR ring network
- FIG. 2 shows the MAC entity structure of each node in a ring direction on the RPR ring network of the present invention
- Figure 3 is a schematic diagram of an RPR ring network using a GE interface
- Figure 4 is a schematic diagram of an RPR ring network with SDH with a rate class of 622 Mbps as the physical layer. Mode for carrying out the invention
- the encapsulation cost of the physical layer always exists, and the encapsulation overhead causes the actual bandwidth that the RPR MAC entity can obtain is smaller than The nominal bandwidth after the speed limit, and because of the randomness of the RPR packet length, the bandwidth loss cannot be accurately calculated. Therefore, there is no way to perform RPR network bandwidth allocation according to the actual bandwidth that the RPR MAC entity can obtain.
- the present invention provides a method for guaranteeing the service level of the RPR ring network.
- the core idea is to accurately control the physical link bandwidth actually occupied by various services on each node of the RPR ring network.
- the nominal bandwidth ensures that all nodes have access to pre-allocated physical link bandwidth.
- an RPR physical packet length is first proposed, which is: in the RPR MAC entity, for each packet that needs to be ringed or a packet that needs to be forwarded, when performing rate limiting
- the packet length is processed according to the RPR physical packet length after the physical layer encapsulation overhead is added, that is, the RPR object is set.
- Packet length RPR file length + physical layer encapsulation overhead.
- the result of rate limiting and rate monitoring using the above RPR physical packet length is equivalent to transferring the physical layer overhead to the RPR MAC entity, increasing the length of each RPR packet by one physical encapsulation overhead, and the physical layer is equivalent to none.
- Any cost so within the RPR MAC entity, the rate limit of the actual physical link bandwidth occupied by various services and the allocated nominal bandwidth can be achieved by performing rate limiting according to the RPR physical packet length of the RPR packet. Therefore, the physical rate of all the upper ring services is precisely limited to the allocated bandwidth, which effectively solves the problem that the forwarding service unexpectedly preempts the bandwidth of the ring service on the downstream node.
- FIG. 2 shows the MAC entity structure of each node in a loop direction on the RPR ring network of the present invention.
- Each node on the RPR ring network includes two looped MAC entities as shown in FIG. 2.
- the ring data from the MAC client of the local node includes services of class A, class B, and class C.
- the service of these three service levels passes through the rate limiting shaper (Shaper) shown in the dotted line of Figure 2.
- SFD policy forwarding scheduling module
- the received data received from the ring corresponding to the MAC entity on the RPR ring network is first input to the detection module, and is separated from the detection module.
- the data that should be received by the local node and the data to be forwarded are outputted, and the data received by the local node is input into the receiving queue, and is output to the MAC client, and the service data of the class A service level in the data to be forwarded is output to the primary forwarding.
- the queue (PTQ) outputs the service data of the class B and the class C service class to be forwarded to the second forwarding queue (STQ), and the PTQ and the STQ respectively output the data to be forwarded for the forwarding and the ring-up service.
- Scheduled SFD SFD will schedule the upper ring traffic after the Shape limit rate and the forwarding traffic from PTQ and STQ And the resulting variety of scheduling the service level traffic flow through this node MAC entity transmits to the other ring RPR network.
- a loopback MAC entity included in the RPR ring network node shown in FIG. 2 further includes: a fairness algorithm bandwidth mediation module that receives fair rate information from the downstream node, According to the congestion state of the node, new fair rate information is generated or the fair rate information received is forwarded, and the upstream node is advertised.
- One of the mediation of the bandwidth algorithm of the fairness algorithm to the congestion is: real-time monitoring of the statistics rate of all non-class AO services forwarded by the node and the upper ring, once the statistical rate of the non-class AO service exceeds the system setting If the bandwidth is not reserved, the congestion information is reported, and the rate of the class C and class B_EIR services in the congested area is adjusted fairly and dynamically.
- the purpose of the above-mentioned fairness algorithm is to correctly determine whether the node is congested, and then start the fair algorithm to adjust the traffic of the upstream node C and B1, so that the non-class AO service does not exceed the non-reserved bandwidth.
- the SFD in FIG. 2 is mainly used for scheduling the upper ring service after the Shaper limited rate and the forwarding service from the PTQ and the STQ.
- the strategy adopted by the scheduling includes:
- PTQ has an absolute priority, that is, as long as there is a class A service to be forwarded, priority is given to forwarding the class A service;
- STQ and STAGE have a relative priority relationship.
- the STAGE data is preferentially scheduled, and once the STQ queue threshold is greater than the overflow threshold, the STQ data is preferentially scheduled. That is to say, when the physical link bandwidth is sufficient and congestion does not occur, both the forwarding data and the upper ring data can be scheduled. Once the data is forwarded and the number of the upper ring is too large, the physical layer is congested.
- the PTQ and STQ data that is, the priority to meet the forwarding data, the upper ring data needs to temporarily stop the service.
- the class A service has higher priority than the class B service, class B.
- the PT/CN2006/001012 service has a higher priority than the class C service.
- the Shaper shown in Figure 2 is rate limited by a token bucket (Token Bucket).
- Token Bucket works as follows: On the one hand, the token is added to the token bucket periodically. On the other hand, each time a packet is sent, the tokens in the token bucket are correspondingly reduced according to the length of the sent packet. If the number of tokens in the token bucket is less than a predetermined low threshold, a flow control indication signal is generated to the client-side MAC entity, and the current packet is stopped, and the token bucket is accumulated in the token bucket. When the number of tokens exceeds the low threshold, the flow control indication signal is revoked, and the client-side MAC entity is allowed to continue to send the current packet again, thereby achieving the purpose of rate control.
- a Token Bucket mainly includes the following six elements: the number of tokens in the token bucket, the number of tokens added in each token period, incSize, the token update period interval, and the token high threshold. High limit, token low threshold and flow control indicator sendX. It is not difficult to infer that by using the Token Bucket for rate limiting, the actual average rate limit is incSize/interval and the maximum burst rate is (high limit-low limit)/interval. It can be seen that the Token Bucket is changed by the single ticket. Different rate controls can be implemented by increasing the number of tokens per the token period or the token update period interval.
- services of different service levels of the upper ring will use different Shaper for rate limiting.
- Different Shapers will have different nominal bandwidths, so that different services will receive different services after rate limiting. bandwidth.
- the class AO and class A1 services use Shaper ShAO and ShAl respectively for rate limiting.
- Shaper ShB and ShF are used for rate limiting.
- class C services Shaper ShF is used for rate limiting.
- the non-class AO service of the upper ring and the non-class AO service forwarded are subject to the speed limit of the Shaper ShD.
- the rate of configuring the Shaper ShD is: physical link bandwidth - (reserved forwarding class AO service bandwidth + reserved upper ring class AO service bandwidth), that is, non-reserved RPR ring network bandwidth. This ensures that the bandwidth of the reserved class AO service can be relatively guaranteed when the link is congested.
- the different Shapers are rate limited according to the RPR physical packet length, and the Shaper performs the following steps in the rate limiting process:
- Al. Determine whether the token update period is reached. If yes, add incSize tokens to the token bucket, but ensure that the number of tokens in the token bucket cannot exceed the preset token high threshold, that is, credit Min (high limit, (credit+incSize));
- Min() represents the minimum value operation
- the physical layer encapsulation overhead length oh is related to the encapsulation technology used. For example, when the RPR packet is transmitted by using the SDH technology, if the general framing process (GFP) encapsulation technology is adopted, the physical layer encapsulation overhead length is Link— oh is 12 bytes. If Link Access Protocol (LAPS) or Advanced Data Link Control (HDLC) encapsulation is used, the physical layer encapsulation overhead length link—oh is 9 bytes. The RPR is transmitted using GE technology. When the message is received, the physical layer encapsulation overhead length link—oh is 20 bytes.
- GFP general framing process
- LAPS Link Access Protocol
- HDLC Advanced Data Link Control
- the number of tokens in the token bucket may be subtracted from the physical layer overhead length link_oh when the current message is about to start to be sent, instead of when the message is sent.
- A4. Determine whether the current credit is less than or equal to the low threshold. If yes, send the flow control indication signal sen dX to the client side MAC entity to stop sending the current packet.
- sendX is equal to 1 to allow the message to be continued.
- 06 001012 Continued transmission, equal to 0 means stop sending the current message.
- the reverse setting is also possible.
- the physical layer of the RPR ring network adopts the GE interface.
- the physical link bandwidth is 1 Gbit/s (Gbps), and the minimum frame interval is 96 bits.
- the physical layer overhead length is 20 bytes.
- the node 2 expects to send a class B-CIR service of 500 megabits per second (Mbps) to the node 4, as described in flow(2, 4) in the figure, assuming that each RPR packet to be sent has a length of 64 words. Section.
- node 3 also expects to send a 500 Mbps class A0 service to node 4, as shown by flow(3, 4) in Figure 3, also assuming Each message to be sent is 64 bytes in length.
- the nominal bandwidth of the Shaper ShB used by the configuration node 2 to limit the class B-CIR service rate is 500 Mbps
- the nominal bandwidth of the Shaper ShD used to limit the rate of the non-class AO service on the upper ring is 500Mbps
- the Shaper ShAO that configures Node 3 to limit the class AO service rate has a nominal rate of 500 Mbps
- the Shaper ShD used to limit the rate of forwarding non-class AO services has a nominal bandwidth of 500 Mbps.
- flow(3,4) is the reserved classAO service bandwidth, which needs absolute protection.
- the Shaper of the node 2 and the node 3 are rate-restricted according to the actual packet length of each RPR packet to be sent, instead of the rate limit according to the RPR physical packet length, the frame interval and the preamble are present due to the data packet transmitted on the physical channel.
- flow(2,4) is sent from the node 2 to the physical channel, since each message will increase the overhead of 20 bytes, the actual occupation of the physical link between the node 2 and the node 4
- the priority of the forwarding service is higher than that of the upper ring service. That is, the RPR ring network node first needs to guarantee the bandwidth of the forwarding service first, so that the node 3 sends the node to the node. 4
- the SHAper of the node 2 and the node 3 have taken the physical layer overhead into consideration in the rate limiting process of each SHAper, and the node 2 sends the flow to the node 4 (2, 4). Regardless of how the RPR packet length changes randomly, the actual physical link bandwidth occupied by the RPR packet will be accurately limited to 500 Mbps, and will not cause the traffic (3, 4) traffic stream bandwidth sent by the node 3 to the node 4, Business flow can be as expected
- each node on the RPR ring network must pass the fairness algorithm shown in Figure 2 in addition to the rate limiting of various services on the ring.
- the bandwidth mediation module allocates the non-reserved bandwidth resources of the RPR ring network to the nodes that compete for the physical link bandwidth resources, and ensures that the statistics rate of the non-class AO services on the RPR ring network does not exceed the unreserved bandwidth. ureservedRate, otherwise, the bandwidth reserved for the class AO service on the RPR ring network cannot be guaranteed.
- the non-reserved bandwidth here is the difference between the physical link bandwidth of the RPR ring network and the bandwidth reserved for the class AO service.
- the fairness algorithm bandwidth reconciliation module shown in FIG. 2 is mainly used to monitor the statistical rate nrXmitRate of all non-class AO services forwarded by the local node and the upper ring in real time, and the non-class AO If the service rate nrXmitRate exceeds the non-reserved bandwidth ureservedRate, the congestion information is reported, and the rate of the upper ring class C and the class B-EIR service in the congested area is adjusted in a fair and dynamic manner, so that the non-class AO service cannot exceed the non-pre- The purpose of leaving bandwidth.
- the fairness algorithm bandwidth mediation module will also calculate all non-class AOs according to the RPR physical packet length of all non-class AO services, in order to accurately calculate the physical link bandwidth actually occupied by all the non-class AO services.
- the statistical rate of the service nrXmitRate, the specific method includes the following two parallel processes B1-B2 and C1 ⁇ C2:
- FIG. 4 is a schematic diagram of an RPR ring network with SDH with a rate class of 622 Mbps as the physical layer. Under this rate level, after the segment overhead and channel overhead are removed, the actual data bandwidth is about 600 Mbps.
- the encapsulation technology uses a GFP package carrying a frame check sequence (FCS), and the physical layer encapsulation overhead is 12 bytes.
- Node 1 and node 2 respectively expect to send a class C service with a bandwidth of 200 Mbps to node 4, as shown by flow(l, 4) and flow(2, 4) of FIG. 4; node 3 expects to send a class AO with a bandwidth of 200 Mbps. Traffic to node 4, as shown in flow (3, 4) of Figure 4.
- the nominal rate of nodes 1 and 2 ShD should be configured to be 400 Mbps, that is, the unreserved bandwidth is 400 Mbps.
- flow(l,4) and flow(2,4) will share this 400Mbps unreserved bandwidth through the adjustment of Node 2's fair algorithm bandwidth mediation module.
- each RPR message transmitted by each node is 64 bytes in length.
- the fairness algorithm bandwidth mediation module of the RPR ring network node is based on the actual length of the RPR packet instead of its RPR physical packet length, at this time, although the node 2 can monitor the flow(l, 4) and flow through real-time monitoring. (2,4)
- the reserved bandwidth of the class AO service flow (3, 4) of the ring on the node 3 is not guaranteed.
- the fairness algorithm bandwidth mediation module of the RPR ring network node is mediated according to the RPR physical packet length of the RPR packet, so when the node 2 performs rate statistics on the non-class AO service
- the physical layer encapsulation cost of each packet is taken into account, so that the sum of the physical link bandwidths when flow(l, 4) and flow(2, 4) reach the physical layer are randomly changed regardless of the length of the PR packet. It will not exceed 400Mbps, so it can effectively guarantee the 200Mbps reserved bandwidth between node 3 and node 4.
- the rate limiting of the RPR physical packet length of the RPR packet and the bandwidth arbitration of the fair algorithm can effectively solve the forwarding service preemption caused by the physical layer encapsulation overhead in the RPR ring network.
- the service bandwidth of the ring node on the downstream node cannot guarantee the defect of reserved bandwidth.
- the rate limiting method described in the present invention is not only applicable to an RPR network, but can be used in other technologies that use a token bucket for bandwidth control.
- a data service processing layer with a network layer above the RPR if the customer configures a B-type service with a guaranteed access rate (CAR) of 20 Mbps and a B-CIR bandwidth of only 20 Mbps, at this time, to ensure the B The bandwidth occupied by the class service is 20 Mbps.
- CAR guaranteed access rate
- the bandwidth occupied by the class service is 20 Mbps.
- the concept of the physical packet length is extended, so that the physical packet length is expressed as the sum of the length of the packet to be sent and the length of the encapsulation overhead of the packet after encapsulation at the next logical layer. .
- the service can be rate-limited according to the physical packet length of a certain service, which mainly includes the following steps: 12, when a token update period of the token bucket is reached, adding a predetermined number of tokens to the token bucket, and limiting the number of tokens in the token bucket to be less than or equal to a preset token high threshold;
- the number of tokens in the current token bucket is subtracted from the number of bytes in the current token bucket.
- the number of tokens in the current token bucket is sent before the current packet to be sent is sent or sent. Subtracting the length of the overhead used by the next logical layer to encapsulate the packet;
- the number of tokens corresponding to the encapsulation overhead length of the next logical layer is used for each sent packet, so that the rate is limited.
- the bandwidth occupied by the next logical layer will be equal to the nominal bandwidth of the token bucket, thus avoiding the bandwidth preemption between various services. happened.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Small-Scale Networks (AREA)
Description
保障数据分组业务服务等级的方法及速率限制方法
技术领域
本发明涉及到面向数据分组业务的数据传送技术, 特别涉及到在弹 性分组环( RPR, Resilient Packet Ring )上保证数据分组业务月良务等级 的方法以及一种基于令牌桶的速率限制方法。 发明背景
RPR环网是一种面向数据分组业务的传送技术, 它将特有的电信级 服务特性与以太网的面向数据业务的高带宽分发、 灵活性及可扩展性有 效结合, 为运营商提供数据优化的带宽管理以及高性价比的多优先级业 务传送解决方案。
图 1显示了现有 RPR环网的结构以及 RPR环网上的节点结构。 如 图 1 所示, RPR环网为互为逆向的双环结构, 即分为内环 (又称为 ringletl )和外环(又称为 ringletO ), 最多可以支持 255个节点互连, 如 图 1中 S0 ~ S254所示, RPR环网上的内、 外环都可以完成数据帧的发 送和接收。 RPR环网上的节点由用户侧实体媒体接入控制客户端(MAC client ), 物理层实体以及媒体接入控制 ( MAC )实体组成, 其中, MAC 实体进一步包含一个 MAC控制实体和两个分别同内、 外环相关联的内 环数据通道实体和外环数据通道实体, 物理层实体可进一步划分成西向 物理层实体和东向物理层实体。 外环从西向物理层实体接收数据, 通过 东向物理层实体发送数据, 内环从东向物理层实体接收数据, 通过西向 物理层实体发送数据, MAC client可以通过内、 外两个环的数据通道收 发数据。
对 RPR环网而言, 对客户侧实体提供不同服务优先级、 并确保协议
要求的服务盾量是必须实现的关键技术之一。 按照 RPR协议标准要求, RPR MAC实体需要为用户侧实体提供 3种服务等级的数据业务: class A、 class B、 class C, 这 3种服务等级业务的特征如下:
class A业务是带宽、抖动鄱有保障的业务。 class A业务在 RPR MAC 实体内部还可以再细分为 class AO和 class A1业务, 其中, class AO业务 的带宽是预先分配的、 并且不允许被其他优先级和其他节点的业务侵 占,该类型业务主要用于承载对时延、抖动有苛刻要求的实时业务; class A1 业务的特点是其带宽、 抖动都要有保障, 但是其带宽不是预留的, 在该业务没有传送需求的情况下, 带宽可以被其他优先级、 其他节点的 业务回收, 从而提高环网带宽利用率, 但是其接入延迟要大于 class A0。
class B业务可在 RPR MAC实体内部进一步细分为 class BO和 class B1业务, 其中, class B0业务为带宽有保障、 抖动和延迟有限度保障的 业务, 又称为在接入承诺速率之内的 class B—CIR业务, 在没有传送需 求情况下, 可以被低优先级的业务回收; class B1业务是速率超过承诺 速率的 class B业务, 为尽力转发的业务, 又称为 class B—EIR业务, 其 带宽、 抖动及延迟均无保障。
class C业务是一种尽力转发的业务,其带宽、抖动及延迟均无保障。 为了保证在 RPR环网上传送的各个服务等级业务的传送带宽, RPR 环网需要对所有上环的、 不同服务等级的业务进行速率限制, 这种速率 限制是实现环网中各种服务等级按照预期分配带宽共享网络资源的必 要前提。 所述速率限制的具体方法为通过精确地获得物理层的链路带 宽,将物理层带宽按照预先的规划,分配到不同的节点及相应的优先级, 保证所有需要带宽保证的服务等级业务的带宽配置之和不大于物理链 路带宽, 这样, 相应的各种业务的带宽都会得到保证。 在各种业务的带 宽得到保证的前提下, 抖动、 延迟可以由 RPR MAC实体的调度机制自
N2006/001012 动得到保证。
但是, 由于 RPR是一种数据链路层技术, 协议标准要求其物理层实 现可以承载到不同的物理层实体上, 目前, 可以选择的有数字同步系列 ( SDH/Sonet )物理层技术和千兆以太网 /万兆以太网 (GE/10GE )物理 层技术等等。 在通过上述物理层技术实现数据传输之前, RPR报文需要 通过特殊的封装或增加一定的开销之后, 才能真正的实现数据传送。 例 如, 对于 SDH技术而言, 它将首先对待传送的数据包(例如 RPR报文) 包封一层长度一定的开销, 之后再映射到 SDH/Sonet的传送时隙上; 而 对于 GE/10GE技术而言, 待传送的数据包(例如 RPR报文)需要进一 步添加前导码 ( Preamble )和帧起始定界符 ( SFD, Start Frame Delimiter ) 之后才能进行传输。 由于在上述封装过程中增加了物理层的开销, 会导 致 RPR MAC实体得到的实际带宽小于其标称带宽。 这种带宽损耗很可 能导致服务等级较低的转发业务, 在链路发生拥塞时, 抢占本节点上环 的服务等级较高业务的带宽, 例如, 在实际的应用中, 由一个节点转发 的 class B一 CIR业务有可能抢占在该节点上环的 class AO业务,或者由一 个节点转发的 class C业务也有可能抢占在该节点上环的 class AO业务, 从而不能实现真正的、 预期的服务等级保障。 发明内容
为了解决上述技术问题, 本发明提供了一种保障数据分组业务服务 等级的方法, 有^避免在链路发生拥塞的情况下, 低服务等级的转发业 务抢占本地高服务等级的上环业务的情况, 实现真正、 预期的业务服务 等级。
本发明的另一个目的在于提供一种速率限制方法, 可以将数据分组 业务的统计速率精确控制在标称带宽之内。
在本发明所述保障数据分组业务服务等级的方法中, 弹性分组环上 的每个节点分别执行以下步骤:
A、 对在本节点上环的不同服务等级的业务, 分别根据各种服务等 级业务的弹性分组环物理包长进行速率限制;
B、 实时才艮据所述各种服务等級业务的弹性分组环物理包长, 对由 本节点转发以及在本节点上环的非 class AO业务进行速率监控, 一旦非 class AO业务的统计速率超过了弹性分组环的非预留带宽, 则上报拥塞 信息, 通过公平算法调解机制调整在拥塞区域内节点上环的非 class AO 业务的速率。
本发明所述根据各种服务等级业务的弹性分组环物理包长进行速率 限制为: 在采用令牌桶进行速率限制的过程中,'每发送完一个弹性分组 环报文或者在每发送一个弹性分组环报文之前, 将当前令牌桶中的令牌 数减去该弹性分组环报文物理层封装开销长度。
本发明所述采用令牌桶进行速率限制包括:
当到达令牌桶的令牌更新周期, 往令牌桶添加预定个数的令牌, 同 时限制令牌桶中令牌的个数小于或等于预先设定的令牌高门限;
在发送弹性分组环报文的过程中, 每发送当前报文的 n个字节, n 小于或等于 256, 将当前令牌桶中的令牌数减去 n;
当当前令牌桶中的令牌数小于或等于预先设定的令牌低门限时, 停 止发送当前的弹性分组环报文。
本发明所述对由本节点转发以及在本节点上环的所有非 class AO业 务进行速率监控为: 根据各种服务等级业务的弹性分组环物理包长, 对 由本节点转发以及在本节点上环的所有非 dass AO业务进行速率统计, 得到所有非 class AO业务的统计速率;
在所述速率统计过程中, 每转发完一个非 class AO业务报文或者在
每转发一个非 class AO业务^ =艮文之前, 将所有非 class AO业务的统计速 率加上所发送报文的物理层封装开销长度;
每发送完一个由本节点上环的非 class AO业务报文或在每发送一个 由本节点上环的非 class AO业务报文之前, 将所有非 class AO业务的统 计速率加上所发送报文的物理层封装开销长度。
在本发明所述的速率限制方法中, 令牌桶根据所发送报文经过下一 逻辑层封装处理后的物理包长进行速率限制, 在所述速率限制过程中, 在当前待发送的报文发送完毕或发送之前, 将当前令牌桶中的令牌数减 去下一逻辑层用于封装该报文的开销长度。
本发明所述速率限制过程包括:
当到达令牌桶的令牌更新周期时, 往令牌桶添加预定个数的令牌, 同时限制所述令牌桶中令牌的个数小于或等于预先设定的令牌高门限; 在发送报文时,将当前令牌桶中的令牌数減去所发送报文的字节数; 当当前令牌桶中的令牌数小于或等于预先设定的令牌低门限时, 停 止发送当前的报文。
由此可以看出, 本发明所述的方法通过根据 RPR报文的 RPR物理 包长对在每个节点上环的业务进行速率限制以及根据 RPR报文的 RPR 物理包长对在每个节点上环或由每个节点转发的所有非 class AO业务进 行速率监控,可以有效地解决 RPR环网中由于物理层封装开销所导致的 转发业务抢占下游节点上环业务带宽, 从而无法保证预留带宽的缺陷。
另外, 本发明所提的技术方案适用随机报文长度, 无须考虑不同的 报文长度所导致的带宽损耗, 按照物理链路带宽分配即可, 保证各个节 点可以得到预期分配物理链路带宽。
附图简要说明
图 1显示了现有 RPR环网的结构以及 RPR环网上的节点结构; 图 2显示了本发明所述 RPR环网上每个节点在一个环向上的 MAC 实体结构;
图 3为采用 GE接口的 RPR环网示意图;
图 4为将速率等级为 622Mbps的 SDH作为物理层的 RPR环网示意 图。 实施本发明的方式
为使发明的目的、 技术方案及优点更加清楚明白, 以下参照附图并 举实施例, 对本发明作进一步详细说明。
由于无论 RPR报文在物理层是由 GE技术承载还是由 SDH/Sonet技 术承载, 其物理层的封装开销总是存在的, 并且这种封装开销会导致 RPR MAC实体可以得到的实际带宽要小于经速度限制后的标称带宽, 同时由于 RPR报文长度的随机性, 这种带宽损耗又无法精确计算出来, 因此没有办法按照 RPR MAC实体可以得到的实际带宽进行 RPR全网带 宽分配。
为此, 本发明提供了一种保障 RPR环网服务等级的方法, 其核心思 想在于精确控制在 RPR环网上每个节点上环的各种业务实际占用的物 理链路带宽不超过为其分配的标称带宽, 从而保证所有节点均可获得预 先分配的物理链路带宽。
为了实现本发明所述的方法, 首先提出一种 RPR物理包长的概念, 其含义在于: 在 RPR MAC实体内, 对于每个需要上环的报文或需要转 发的报文, 当进行速率限制以及速率监控的时候, 都将其报文长度按照 增加了物理层封装开销之后的 RPR物理包长进行处理, 即设定 RPR物
理包长 =RPR 文长度 +物理层封装开销。
使用上述 RPR物理包长进行速率限制及速率监控的结果相当于将 物理层的开销转移到 RPR MAC实体内部,将每个 RPR报文的长度增加 一个物理封装开销,而物理层则等效于没有任何开销,因此在 RPR MAC 实体内部, 只要按照 RPR报文的 RPR物理包长进行速率限制, 即可达 到各种业务所占用的实际物理链路带宽同所分配的标称带宽精确匹配 的效果, 从而使所有的上环业务物理速率精确限制在分配的带宽之内, 有效地解决转发业务非预期抢占下游节点上环业务带宽问题。
图 2显示了本发明所述 RPR环网上每个节点在一个环向上的 MAC 实体结构。 RPR环网上的每个节点包括如图 2所示的两个环向的 MAC 实体。 如图 2所示,从本节点 MAC Client上环数据包括 class A、 class B 和 class C三种服务等级的业务,这三种服务等级的业务经过图 2虚线框 所示速率限制整形器 (Shaper )进行速率限制后输出到用于对转发以及 上环业务进行调度的策略转发调度模块 ( SFD ); 从 RPR环网上本 MAC 实体对应的环向上接收到的接收数据首先输入到检测模块, 从中分离出 本节点应当接收的数据以及待转发的数据, 并将本节点接收的数据输入 到接收队列中, 准备输出给 MAC Client, 将待转发的数据中的 class A 服务等级的业务数据输出到主转发队列 (PTQ ), 而将待转发的 class B 和 class C服务等级的业务数据输出到第二转发队列 (STQ ), PTQ以及 STQ 将待转发的数据分别输出到用于对转发以及上环业务进行调度的 SFD; SFD将对经过 Shaper限制速率后的上环业务以及来自 PTQ和 STQ 的转发业务进行调度, 并将调度后的得到的各种服务等级的业务流通过 本节点另一个环向的 MAC实体发送到 RPR环网上。
图 2所示的 RPR环网节点所包含的一个环向的 MAC实体还将进一 步包括: 公平算法带宽调解模块, 接收来自下游节点的公平速率信息,
根据本节点拥塞状态, 产生新的公平速率信息或转发接收的公平速率信 息, 继续向上游节点通告。 所述公平算法带宽调解模块对拥塞的判断调 解之一是: 实时监测本节点转发以及上环的所有非 class AO业务的统计 速率, 一旦所述非 class AO业务的统计速率超过了系统设定的非预留的 带宽, 则上报拥塞信息, 公平、 动态的调整拥塞区域的上环 class C和 class B_EIR业务的速率。 上述公平算法带宽调解也即速率监控的目的 是, 正确判断节点是否拥塞, 进而启动公平算法调节上游节点 C和 B1 业务流量, 达到非 classAO业务不超过非预留带宽的目的。
图 2中的 SFD主要用于对经过 Shaper限制速率后的上环业务以及 来自 PTQ和 STQ的转发业务进行调度, 在本发明的优选实施例中所述 调度采用的策略包括:
1. 上环的 class A、 class B以及 class C 3种业务合路后, 将形成一 个逻辑上的数据队列 STAGE, 该 STAGE队列同转发的数据竟争物理链 路带宽。
2. 转发队列 PTQ、 STQ和 STAGE队列对物理链路带宽的占用优先 级顺序为:
PTQ具有绝对优先级, 即只要有转发的 class A业务,要优先保证转 发 class A业务;
STQ和 STAGE具有相对优先级关系, 当 STQ的队列门限小于一预 先设定的溢出门限时, 优先调度 STAGE数据, 而一旦 STQ队列门限大 于溢出门限, 则优先调度 STQ数据。 也就是说, 当物理链路带宽足够, 未发生拥塞时, 转发数据和上环数据都能得到调度, 一旦转发数据和上 环数椐由于流量太大造成物理层拥塞时, 则要优先保证转发的 PTQ和 STQ数据, 即优先满足转发数据, 上环数据需要暂时停止服务。
3.对于上环的业务, class A业务的优先级高于 class B业务, class B
P T/CN2006/001012 业务的优先级高于 class C业务。
在本发明的优选实施例中,图 2所示的 Shaper是通过令牌桶(Token Bucket )来实现速率限制的。 Token Bucket的工作原理为: 一方面周期 性地往令牌桶中增加令牌, 另一方面每发送一个报文, 都要根据所发送 艮文的长度相应地减少令牌桶中的令牌个数, 在发送报文的过程中, 若 令牌桶中令牌数小于一个预定的低门限, 则产生一个流控指示信号给客 户侧 MAC实体, 停止发送当前报文, 待令牌桶中累积的令牌数目超过 所述低门限时, 撤消所述流控指示信号, 再次允许客户侧 MAC实体继 续发送当前报文, 从而实现速率控制的目的。
按照 RPR标准建议, 一个 Token Bucket主要包括以下 6个要素:令 牌桶中的令牌个数 credit, 每个令牌周期内增加的令牌个数 incSize、 令 牌更新周期 interval、 令牌高门限 high limit, 令牌低门限 low limit以及 流控指示信号 sendX。 不难推断,通过使用 Token Bucket进行速率限制, 实际得到的平均限制速率为 incSize/interval , 最大突发速率为(high limit-low limit)/ interval 由此可以看出, 通过筒单改变 Token Bucket在 每个令牌周期内增加的令牌个数 incSize或令牌更新周期 interval, 就可 以实现不同的速率控制。
如图 2所示, 对于上环的不同服务等级的业务将使用不同的 Shaper 进行速率限制, 其中, 不同的 Shaper将具有不同的标称带宽, 从而使得 不同的业务经过速率限制后得到不同的业务带宽。 例如, 对于 class AO 和 class A1业务分别使用 Shaper ShAO和 ShAl进行速率限制,对于 class B0和 class Bl业务分别使用 Shaper ShB和 ShF进行速率限制,对于 class C业务将使用 Shaper ShF进行速率限制。 通过使用不同的 Shaper, 不同 服务等级的业务将得到不同速率的带宽。 除此之外, 上环的非 class AO 业务和转发的非 class AO业务还要受到 Shaper ShD的速率限制,在本发
2 明的优选实施例中, 配置 Shaper ShD限制的速率为: 物理链路带宽- (预 留的转发 class AO业务带宽 +预留的上环 class AO业务带宽), 即 RPR环 网的非预留带宽。 这样可以确保在链路发生拥塞时, 预留的 class AO业 务的带宽也能够得到相对保障。
在本发明的优选实施例中, 上述不同的 Shaper均是根据 RPR物理 包长进行速率限制的, 所述 Shaper在速率限制过程中执行以下步骤:
Al、 判断是否到达令牌更新周期, 如果是, 则往令牌桶添加 incSize 个令牌, 但是保证令牌桶中令牌的个数不得超过预先设定的令牌高门 限, 即令 credit=Min(high limit,(credit+incSize));
其中, 函数 Min()表示取最小值运算;
A2、 判断是否有报文发送, 如果是, 则每发送 n个字节(n < 256 ), 将减去所发送的字节数个令牌, 即令 credit=credit-n;
A3、 判断当前报文是否发送完毕, 如果报文已经发送完毕, 则将令 牌数减去物理层封装开销长度 link— oh, 即令 credit=credit-link— oh;
该步骤所述的物理层封装开销长度 link— oh与所采用的封装技术有 关,例如在采用 SDH技术传输 RPR报文时,若采用通用成帧过程( GFP ) 封装技术, 则物理层封装开销长度 link— oh为 12字节, 若采用链路接入 协议 ( LAPS )或高级数据链路控制 (HDLC )封装技术, 则物理层封装 开销长度 link— oh为 9字节; 在采用 GE技术传输 RPR报文时, 物理层 封装开销长度 link— oh为 20字节。
在该步骤中, 也可以在即将开始发送当前的报文时, 而非在才艮文发 送完毕时, 将令牌桶中的令牌数减去物理层开销长度 link— oh;
A4、 判断当前 credit是否小于或等于低门限, 如果是, 则发送流控 指示信号 sendX给客户侧 MAC实体, 停止发送当前报文。
根据本发明的优选实施例, 可以设置 sendX等于 1表示允许报文继
06 001012 续发送,等于 0表示停止发送当前报文, 当然,反过来设置也是可以的。
对于根据各种非 class AO业务的 RPR物理包长,对上环的非 class AO 业务和转发的非 Class AO业务进行速率限制的 Shaper ShD来讲,在执行 上述步骤 A1 - A4的过程中,它还将进一步在每转发 n个字节( n《 256 ) 时,将令牌桶中的令牌数减去所转发的字节数; 在每转发完一个 RPR报 文或在转发 RP 报文之前将令牌桶中的令牌数减去该 RPR报文的物理 层封装开销长度; 并且在监测到令牌桶中的令牌数小于或等于所述令牌 低门限时,仅仅停止发送由本节点上环的 RPR报文, 而不停止转发在本 节点转发的 RPR报文。
从上面的速率限制流程来看,在相应的 Shaper进行速率限制的过程 中,对于每个上环的 RPR报文均将多用去一个相当于物理层封装开销长 度 link— oh的令牌数,则经过该 Shaper限制速率的 RPR报文达到物理层 时,该 RPR报文加上每个报文的物理封装开销后所占用的实际物理链路 带宽将刚好等于该 Shaper所配置的标称带宽, 从而避免了每个 RPR环 网节点非预期过多占用物理链路带宽情况的发生,精确控制 RPR环网上 每个节点的上环速率不超过其分配的物理链路带宽, 从而保证所有节点 获得预先分配的物理链路带宽。
下面将结合图 3 , 通过一个具体示例详细说明本发明优选实施例所 述的速率限制方法。
如图 3所示, RPR环网中有 4个节点, RPR环网的物理层采用 GE 接口, 物理链路带宽为 1吉比特 /秒 ( Gbps ), 最小的帧间隔为 96比特, 可以等效为物理层开销长度为 20字节。 其中, 节点 2期望发送 500兆 比特 /秒 ( Mbps )的 class B— CIR业务到节点 4, 如图中的 flow(2,4)所述, 假定待发送的每个 RPR报文长度为 64字节。 另外, 节点 3也期望发送 500 Mbps的 class A0业务到节点 4, 如图 3中的 flow(3,4)所示, 也假定
待发送的每个报文长度为 64字节。
按照上述带宽要求, 根据 RPR协议, 将配置节点 2用于限制 class B— CIR业务速率的 Shaper ShB的标称带宽为 500Mbps;用于限制上环非 class AO业务速率的 Shaper ShD的标称带宽为 500Mbps; 将配置节点 3 用于限制 class AO业务速率的 Shaper ShAO的标称速率为 500Mbps, 用 于限制转发非 class AO业务速率的 Shaper ShD的标称带宽为 500Mbps。 其中, flow(3,4)为预留的 classAO业务带宽, 需要绝对保障。
若上述节点 2和节点 3的 Shaper是按照各个待发送 RPR报文的实 际包长, 而不是按照其 RPR物理包长进行速率限制,则由于在物理信道 上发送地数据包存在帧间隔、 前导码以及 SFD, flow(2,4)从节点 2发送 到物理信道上之后, 由于每个报文都将增加 20字节的开销, 这样, 在 节点 2 和节点 4 之间对物理链路实际的占用带宽变为 500 X (64+20)/64=656Mbpso 而 flow(2,4)是需要节点 3的转发才能到达节点 4 的, 即 flow(2,4)对节点 3来讲为转发业务。 从前面描述的调度策略可以 知道,转发业务在链路发生拥塞时,优先级是要高于上环业务的,即 RPR 环网节点首先要优先保证转发业务的带宽, 这样, 节点 3发送到节点 4 的上环业务, 即 flow(3,4)可以得到的物理链路带宽只剩 1000-656=344 Mbps, 远小于分配预期, 因而导致预留给节点 3上环的 class AO业务 flow(3,4)的带宽无法得到保障。
然而,根据本发明优选实施例所述的方法,节点 2和节点 3的 Shaper 于各个 Shaper在速率限制过程中已经将物理层开销考虑在内, 则节点 2 发送到节点 4的 flow(2,4)无论 RPR报文长度如何随机变化,其实际占用 的物理链路带宽将精确的限制在 500Mbps以内,不会造成对节点 3发送 到节点 4的 flow(3,4)业务流带宽的抢占, 两条业务流均可以按照预期的
12
01012 分配合理地占用物理链路带宽。
为了满足 RPR环网上各种业务带宽需求, 实现不同等级的服务质 量, RPR环网上的每个节点除了要对上环的各种业务进行速率限制之 外,还要通过图 2所示的公平算法带宽调解模块公平、动态地将 RPR环 网中非预留的带宽资源分配给竟争物理链路带宽资源的各个节点, 保证 RPR 环网上的非 class AO 业务的统计速率不超过非预留的带宽 ureservedRate, 否则, 为 RPR环网上 class AO业务预留的带宽就无法得 到保障了。在这里所述非预留带宽为 RPR环网的物理链路带宽与为 class AO业务预留的带宽之差。
在本发明的优选实施例中, 图 2所示的公平算法带宽调解模块将主 要用于实时监测本节点转发的以及上环的所有非 class AO业务的统计速 率 nrXmitRate, —旦所述非 class AO业务的统计速率 nrXmitRate超过了 所述非预留的带宽 ureservedRate, 则上报拥塞信息, 公平、 动态的调整 拥塞区域的上环 class C和 classB— EIR业务的速率,实现非 class AO业务 不能超过非预留带宽的目的。
为了精确统计出所有非 class AO业务实际占用的物理链路带宽, 本 发明优选实施例所述的公平算法带宽调解模块也将根据所有非 class AO 业务的 RPR 物理包长统计所述所有非 class AO 业务的统计速率 nrXmitRate, 具体方法包括以下两个并行的过程 B1-B2和 C1~C2:
Bl、 判断当前从本节点上环的 RPR报文是否为 class AO业务报文, 若不是 class AO业务报文,则每发送 n个有效字节,则令所有非 class AO 业务的统计速率 nrXmitRate力 p n, 即令 nrXmitRate=nrXmitRate+n, 然后 执行 B2; 否则, 直接返回本步骤 B1 ;
B2、 在当前发送的非 class AO业务报文发送完毕后 (或发送之前), 令所有非 class AO业务的统速率 nrXmitRate加上物理层封装开销长度
link— oh, 即令 nrXmitRate=nrXmitRate+link— oh; 否则, 返回步骤 Bl;
Cl、 判断当前转发的 RPR报文是否为 class AO业务报文, 若不是 class AO业务报文, 则每发送 n个有效字节, 则令所有非 class AO业务 的统计速率 nrXmitRate加 n, 即令 nrXmitRate=nrXmitRate+n , 然后执行 C2; 否则, 直接返回本步骤 C1;
C2、 在当前转发的非 class AO业务报文转发完毕后 (或发送之前), 令所有非 class AO业务的统计速率 nrXmitRate加上物理层封装开销长度 link— oh, 即令 nrXmitRate=nrXmitRate+link— oh; 否则, 返回步 。1。
下面将结合图 4, 通过另一具体示例详细说明本发明优选实施例所 述的公平算法带宽调解方法。
图 4为一个将速率等级为 622Mbps的 SDH作为物理层的 RPR环网 示意图。 在这种速率等级之下, 将段开销和通道开销去除后, 实际的数 据带宽约为 600Mbps, 封装技术采用携带帧校验序列 (FCS )的 GFP封 装, 物理层封装开销长度为 12字节。 其中, 节点 1和节点 2分别期望 发送带宽为 200Mbps的 class C业务到节点 4, 如图 4的 flow(l,4)和 flow(2,4)所示;节点 3期望发送带宽为 200Mbps的 classAO业务到节点 4, 如图 4的 flow(3,4)所示。
按照上述速率要求,根据 RPR协议,应当配置节点 1和 2 ShD标称 速率为 400Mbps,即非预留带宽为 400Mbps。在实际的应用中, flow(l,4) 和 flow(2,4)将通过节点 2 的公平算法带宽调解模块的调节共享这 400Mbps非预留带宽。为了便于描述假定每个节点传送的每个 RPR报文 长度均为 64字节。
若 RPR环网上节点的公平算法带宽调解模块是根据 RPR报文的实 际长度, 而非其 RPR物理包长进行调解的, 此时, 虽然节点 2通过实时 监控可以将 flow(l,4)和 flow(2,4)的上环业务带宽之和控制在 400Mbps之
内, 但是, 这两条数据流实际占用的物理链路带宽将为 400 X
由于 flow(l,4)和 flow(2,4)需要经过节点 3的转发 才能到达节点 4, 因此节点 3 可以获得的物理链路带宽就仅剩下: 600-475=125Mbps , 远小于预期分配给节点 3 上环的 class AO 业务 flow(3,4)的预留带宽, 导致预留带宽无法得到保障。
然而, 根据本发明优选实施例所述的方法, RPR环网上节点的公平 算法带宽调解模块是根据 RPR报文的 RPR物理包长进行调解的 , 因而 当节点 2对非 class AO业务进行速率统计时, 将每个报文的物理层封装 开销考虑在内, 使得无论 PR报文长度如何随机变化, flow(l,4)和 flow(2,4)到达物理层时的物理链路带宽之和均不会超过 400Mbps, 因此 可以有效的保证节点 3到节点 4之间 200Mbps预留带宽。
通过上述本发明的优选实施例可以看出,通过 ^据 RPR报文的 RPR 物理包长进行速率限制以及公平算法带宽调解可以有效地解决 RPR环 网中由于物理层封装开销所导致的转发业务抢占下游节点上环业务带 宽, 从而无法保证预留带宽的缺陷。
另外, 熟悉本领域的技术人员可以理解, 本发明所述的速率限制方 法不仅仅适用于 RPR网络,在其它利用令牌桶进行带宽控制的技术中都 可以使用。 例如, 对于网络层次处于 RPR之上的数据业务处理层, 若客 户配置某 B类业务的承诺接入速率( CAR )为 20Mbps, 而 B—CIR带宽 仅为 20Mbps, 此时, 如要保证该 B类业务所占有的带宽在 20Mbps之 为此, 将物理包长的概念进行扩展, 令物理包长表示为待发送报文 的长度与该报文在下一逻辑层进行封装后的封装开销长度之和。 这样 , 就可以根据某种业务的物理包长来对该业务进行速率限制了, 主要包括 以下步骤:
12 当到达令牌桶的令牌更新周期时, 往令牌桶添加预定个数的令牌 , 同时限制所述令牌桶中令牌的个数小于或等于预先设定的令牌高门限; 在发送报文时,将当前令牌桶中的令牌数减去所发送报文的字节数; 在当前待发送的报文发送完毕或发送之前, 将当前令牌桶中的令牌 数减去下一逻辑层用于封装该报文的开销长度;
当当前令牌桶中的令牌数小于或等于预先设定的令牌低门限时, 停 止发送当前的报文。
通过上述方法可以看出, 在上述速率限制过程中, 对于每个所发送 的报文均将多用去个数相当于其下一逻辑层的封装开销长度的令牌数, 这样, 经过上述限制速率的报文到达下一逻辑层时, 该 ^艮文加上下一逻 辑层的封装开销后所占用的带宽将刚好等于该令牌桶的标称带宽, 从而 避免了各种业务之间带宽抢占情况的发生。
Claims
1、 一种保障数据分組业务服务等级的方法, 其特征在于, 弹性分组 环上的每个节点分别执行以下步驟:
A、 对在本节点上环的不同服务等级的业务, 分别才艮据各种服务等 级业务的弹性分组环物理包长进行速率限制;
B、 实时根据所述各种服务等级业务的弹性分组环物理包长, 对由 本节点转发以及在本节点上环的非 class AO 业务进行速率监控, 当非 class AO业务的统计速率超过弹性分组环的非预留带宽时, 上艮拥塞信 息, 调整在拥塞区域内节点上环的非 class AO业务速率。
2、 如权利要求 1所述的方法, 其特征在于, 所述弹性分组环物理包 长为弹性分组环报文长度与物理层封装开销长度之和。
3、 如权利要求 1所述的方法, 其特征在于, 所述根据各种服务等级 业务的弹性分组环物理包长进行速率限制为: 在采用令牌桶进行速率限 制的过程中, 每发送完一个弹性分组环报文, 就将当前令牌桶中的令牌 数减去该弹性分组环 4艮文物理层封装开销长度。
4、 如权利要求 1所述的方法, 其特征在于, 所述根据各种服务等级 业务的弹性分组环物理包长进行速率限制为: 在采用令牌桶进行速率限 制的过程中, 在每发送一个弹性分组环报文之前, 将当前令牌桶中的令 牌数减去该弹性分组环报文物理层封装开销长度。
5、如权利要求 3或 4所述的方法,所述采用令牌桶进行速率限制过 程包括:
当到达令牌桶的令牌更新周期, 往令牌桶添加预定个数的令牌, 同 时限制令牌桶中令牌的个数小于或等于预先设定的令牌高门限;
在发送弹性分组环报文的过程中, 每发送当前报文的 n个字节, n
小于或等于 256, 将当前令牌桶中的令牌数减去 n;
当当前令牌桶中的令牌数小于或等于预先设定的令牌低门限时, 停 止发送当前的弹性分組环报文。
6、 如权利要求 1所述的方法, 其特征在于, 所述对由本节点转发以 及在本节点上环的所有非 class AO业务进行速率监控为: 根据各种服务 等级业务的弹性分组环物理包长, 对由本节点转发以及在本节点上环的 所有非 class AO业务进行速率统计, 得到所有非 class AO业务的统计速 率;
在所述速率统计过程中, 每转发完一个非 class AO业务报文, 将所 有非 class AO业务的统计速率加上所发送艮文的物理层封装开销长度; 每发送完一个由本节点上环的非 class AO业务报文, 将所有非 class AO业务的统计速率加上所发送报文的物理层封装开销长度。
7、如权利要求 1所述的方法, 其特征在于, 所述对由本节点转发以 及在本节点上环的所有非 class AO业务进行速率监控为: 根据各种服务 等级业务的弹性分组环物理包长, 对由本节点转发以及在本节点上环的 所有非 class AO业务进行速率统计, 得到所有非 class AO业务的统计速 率;
在所述速率统计过程中, 在每转发一个非 class AO业务报文之前,
度;
在每发送一个由本节点上环的非 class AO业务报文之前, 将所有非 class AO业务的统计速率加上所发送报文的物理层封装开销长度。
8、如权利要求 6或 7所述的方法,其特征在于,所述速率统计包括: 在当前转发的弹性分組环报文为非 class AO业务报文时, 每发送 n 个有效字节, n小于或等于 256,将所有非 class AO业务的统计速率加 n;
在当前从本节点上环的弹性分组环报文为非 class AO业务报文时, 每发送 n个有效字节, 将所有非 class AO业务的统计速率加 n。
9、 如权利要求 1所述的方法, 其特征在于, 步骤 B所述调整在拥 塞区域内节点上环的非 class AO业务速率包括: 调整在拥塞区域内节点 上环的 class C和 class B— EIR业务速率。
10、 如权利要求 1所述的方法, 其特征在于, 所述方法在步骤 A之 后进一步包括: 根据各种服务等級业务的弹性分组环物理包长, 对在本 节点的上环以及由本节点转发的所有非 class AO业务进行进一步的速率 限制, 将所有非 class AO业务的物理链路带宽之和限制在弹性分组环网 的非预留带宽之内。
11、 如权利要求 10所述的方法, 其特征在于, 所述根据各种服务等 級业务的弹性分组环物理包长进行速率限制具体包括: 在采用令牌桶进 行速率限制的过程中, 每发送或转发完一个非 class AO业务报文, 或者 在每发送或转发一个非 class AO业务报文之前, 将当前令牌桶中的令牌 数减去该非 class AO业务报文物理层封装开销长度, 并且在监测到令牌 桶中的令牌数小于或等于所述令牌低门限时, 停止发送由本节点上环的 非 class AO业务报文。
12、 一种速率限制方法, 其特征在于, 令牌桶根据所发送报文经过 下一逻辑层封装处理后的物埋包长进行速率限制, 在所述速率限制过程 中, 在当前待发送的报文发送完毕或发送之前, 将当前令牌桶中的令牌 数减去下一逻辑层用于封装该报文的开销长度。
13、如权利要求 12所述的方法, 其特征在于, 所述物理包长为当前 待发送报文长度与该报文在下一逻辑层的封装开销长度之和。
14、如权利要求 12所述的方法, 其特征在于, 所述速率限制过程包 括:
当到达令牌桶的令牌更新周期时, 往令牌桶添加预定个数的令牌, 同时限制所述令牌桶中令牌的个数小于或等于预先设定的令牌高门限; 在发送报文时,将当前令牌桶中的令牌数减去所发送报文的字节数; 当当前令牌桶中的令牌数小于或等于预先设定的令牌低门限时, 停 止发送当前的报文。
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP06741900A EP1814265B1 (en) | 2005-10-31 | 2006-05-17 | A method for guaranteeing classificaiton of service of the packet traffic and the method of rate restriction |
AT06741900T ATE500667T1 (de) | 2005-10-31 | 2006-05-17 | Verfahren zum garantieren der klassifizierung des paketverkehrsdienstes und des ratenbegrenzungsverfahrens |
US11/568,311 US7684348B2 (en) | 2005-10-31 | 2006-05-17 | Method for ensuring service class of packet service and method of rate limitation |
DE602006020400T DE602006020400D1 (de) | 2005-10-31 | 2006-05-17 | Verfahren zum garantieren der klassifizierung des paketverkehrsdienstes und des ratenbegrenzungsverfahrens |
CNB2006800118990A CN100558069C (zh) | 2005-10-31 | 2006-05-17 | 保障数据分组业务服务等级的方法及速率限制方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200510117076.3 | 2005-10-31 | ||
CNB2005101170763A CN100433718C (zh) | 2005-10-31 | 2005-10-31 | 保障数据分组业务服务等级的方法及速率限制方法 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2007051374A1 true WO2007051374A1 (en) | 2007-05-10 |
WO2007051374A8 WO2007051374A8 (fr) | 2007-07-19 |
Family
ID=37133711
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2006/001012 WO2007051374A1 (en) | 2005-10-31 | 2006-05-17 | A method for guaranteeing classification of service of the packet traffic and the method of rate restriction |
Country Status (6)
Country | Link |
---|---|
US (1) | US7684348B2 (zh) |
EP (1) | EP1814265B1 (zh) |
CN (2) | CN100433718C (zh) |
AT (1) | ATE500667T1 (zh) |
DE (1) | DE602006020400D1 (zh) |
WO (1) | WO2007051374A1 (zh) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7880028B2 (en) | 2006-07-14 | 2011-02-01 | Invista North America S.A R.L. | Process for making 3-pentenenitrile by hydrocyanation of butadiene |
US7897801B2 (en) | 2003-05-12 | 2011-03-01 | Invista North America S.A R.L. | Process for the preparation of dinitriles |
US7919646B2 (en) | 2006-07-14 | 2011-04-05 | Invista North America S.A R.L. | Hydrocyanation of 2-pentenenitrile |
US7973174B2 (en) | 2005-10-18 | 2011-07-05 | Invista North America S.A.R.L. | Process of making 3-aminopentanenitrile |
US7977502B2 (en) | 2008-01-15 | 2011-07-12 | Invista North America S.A R.L. | Process for making and refining 3-pentenenitrile, and for refining 2-methyl-3-butenenitrile |
US8088943B2 (en) | 2008-01-15 | 2012-01-03 | Invista North America S.A R.L. | Hydrocyanation of pentenenitriles |
US8101790B2 (en) | 2007-06-13 | 2012-01-24 | Invista North America S.A.R.L. | Process for improving adiponitrile quality |
US8178711B2 (en) | 2006-03-17 | 2012-05-15 | Invista North America S.A R.L. | Method for the purification of triorganophosphites by treatment with a basic additive |
US8247621B2 (en) | 2008-10-14 | 2012-08-21 | Invista North America S.A.R.L. | Process for making 2-secondary-alkyl-4,5-di-(normal-alkyl)phenols |
US8338636B2 (en) | 2009-08-07 | 2012-12-25 | Invista North America S.A R.L. | Hydrogenation and esterification to form diesters |
US8913501B2 (en) | 2012-08-16 | 2014-12-16 | International Business Machines Corporation | Efficient urgency-aware rate control scheme for multiple bounded flows |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5056341B2 (ja) * | 2006-11-07 | 2012-10-24 | 富士通株式会社 | 通信中継装置、通信中継方法および通信中継処理プログラム |
CN101687658B (zh) | 2007-05-14 | 2013-07-24 | 因温斯特技术公司 | 高效反应器和方法 |
FR2934447A1 (fr) * | 2008-07-23 | 2010-01-29 | France Telecom | Procede de communication entre une pluralite de noeuds, les noeuds etant organises suivant un anneau |
KR101233915B1 (ko) | 2008-12-25 | 2013-02-15 | 미쓰비시덴키 가부시키가이샤 | 통신 관리 장치, 통신 장치 및 통신 방법 |
CN101951339B (zh) * | 2010-08-24 | 2012-08-15 | 北京星网锐捷网络技术有限公司 | 报文限速方法、装置及网络设备 |
CN102111340B (zh) * | 2011-03-30 | 2013-01-02 | 北京星网锐捷网络技术有限公司 | 带宽的限制方法、装置及网络设备 |
US9055467B2 (en) * | 2013-01-14 | 2015-06-09 | Dell Products L.P. | Sender device based pause system |
US9608935B2 (en) * | 2014-09-08 | 2017-03-28 | Qualcomm Technologies, Inc. | Tunneling within a network-on-chip topology |
US9654483B1 (en) * | 2014-12-23 | 2017-05-16 | Amazon Technologies, Inc. | Network communication rate limiter |
US9935885B1 (en) | 2016-03-15 | 2018-04-03 | Juniper Networks, Inc. | Managing flow table entries for express packet processing based on packet priority or quality of service |
CN111510394B (zh) * | 2019-01-31 | 2022-04-12 | 华为技术有限公司 | 一种报文调度方法、相关设备及计算机存储介质 |
CN112953848B (zh) * | 2021-03-12 | 2023-05-02 | 西安电子科技大学 | 一种基于严格优先级的流量监管方法、系统及设备 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2002073903A1 (en) * | 2001-03-12 | 2002-09-19 | Luminous Networks, Inc. | Bandwidth reservation reuse in dynamically allocated ring protection and restoration technique |
CN1601963A (zh) * | 2003-09-26 | 2005-03-30 | 华为技术有限公司 | 基于令牌桶算法的接入速率限制方法 |
US20050100031A1 (en) * | 2003-11-11 | 2005-05-12 | Byung-Gu Choe | Allocating bandwidth using resilient packet ring (RPR) fairness mechanism |
CN1622530A (zh) * | 2003-11-25 | 2005-06-01 | 中兴通讯股份有限公司 | 一种基于业务分类的端口速率控制方法 |
Family Cites Families (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6687247B1 (en) * | 1999-10-27 | 2004-02-03 | Cisco Technology, Inc. | Architecture for high speed class of service enabled linecard |
US6816494B1 (en) * | 2000-07-20 | 2004-11-09 | Nortel Networks Limited | Method and apparatus for distributed fairness algorithm for dynamic bandwidth allocation on a ring |
WO2003025709A2 (en) * | 2001-09-19 | 2003-03-27 | Bay Microsystems, Inc. | Vertical instruction and data processing in a network processor architecture |
US20030163593A1 (en) * | 2002-02-25 | 2003-08-28 | William March Rice University | Method and system for implementing a fair, high-performance protocol for resilient packet ring networks |
CN1194509C (zh) * | 2002-06-12 | 2005-03-23 | 华为技术有限公司 | 一种控制路由器接口报文发送速率的方法 |
AU2002313444A1 (en) * | 2002-07-17 | 2004-02-02 | Wuhan Fiberhome Networks Co., Ltd. | Multiple service ring with capabilities of transmitting and switching data, video and voice |
US20040032826A1 (en) * | 2002-08-02 | 2004-02-19 | Kamakshi Sridhar | System and method for increasing fairness in packet ring networks |
US7369504B2 (en) * | 2002-12-12 | 2008-05-06 | Alcatel Canada, Inc. | Bandwidth management of resilient packet ring network |
CN1266895C (zh) * | 2003-03-11 | 2006-07-26 | 华为技术有限公司 | 动态调整内嵌式弹性分组环环网带宽的方法 |
US7590130B2 (en) * | 2004-12-22 | 2009-09-15 | Exar Corporation | Communications system with first and second scan tables |
-
2005
- 2005-10-31 CN CNB2005101170763A patent/CN100433718C/zh not_active Expired - Fee Related
-
2006
- 2006-05-17 US US11/568,311 patent/US7684348B2/en active Active
- 2006-05-17 AT AT06741900T patent/ATE500667T1/de not_active IP Right Cessation
- 2006-05-17 EP EP06741900A patent/EP1814265B1/en not_active Not-in-force
- 2006-05-17 CN CNB2006800118990A patent/CN100558069C/zh not_active Expired - Fee Related
- 2006-05-17 DE DE602006020400T patent/DE602006020400D1/de active Active
- 2006-05-17 WO PCT/CN2006/001012 patent/WO2007051374A1/zh active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2002073903A1 (en) * | 2001-03-12 | 2002-09-19 | Luminous Networks, Inc. | Bandwidth reservation reuse in dynamically allocated ring protection and restoration technique |
CN1601963A (zh) * | 2003-09-26 | 2005-03-30 | 华为技术有限公司 | 基于令牌桶算法的接入速率限制方法 |
US20050100031A1 (en) * | 2003-11-11 | 2005-05-12 | Byung-Gu Choe | Allocating bandwidth using resilient packet ring (RPR) fairness mechanism |
CN1622530A (zh) * | 2003-11-25 | 2005-06-01 | 中兴通讯股份有限公司 | 一种基于业务分类的端口速率控制方法 |
Non-Patent Citations (1)
Title |
---|
CHENG S.: "Resilient Packet Ring technology", ZTE COMMUNICATION TECHNOLOGY, no. 5, May 2005 (2005-05-01), pages 48 - 51, XP008080503 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7897801B2 (en) | 2003-05-12 | 2011-03-01 | Invista North America S.A R.L. | Process for the preparation of dinitriles |
US7973174B2 (en) | 2005-10-18 | 2011-07-05 | Invista North America S.A.R.L. | Process of making 3-aminopentanenitrile |
US8178711B2 (en) | 2006-03-17 | 2012-05-15 | Invista North America S.A R.L. | Method for the purification of triorganophosphites by treatment with a basic additive |
US7880028B2 (en) | 2006-07-14 | 2011-02-01 | Invista North America S.A R.L. | Process for making 3-pentenenitrile by hydrocyanation of butadiene |
US7919646B2 (en) | 2006-07-14 | 2011-04-05 | Invista North America S.A R.L. | Hydrocyanation of 2-pentenenitrile |
US8394981B2 (en) | 2006-07-14 | 2013-03-12 | Invista North America S.A R.L. | Hydrocyanation of 2-pentenenitrile |
US8101790B2 (en) | 2007-06-13 | 2012-01-24 | Invista North America S.A.R.L. | Process for improving adiponitrile quality |
US7977502B2 (en) | 2008-01-15 | 2011-07-12 | Invista North America S.A R.L. | Process for making and refining 3-pentenenitrile, and for refining 2-methyl-3-butenenitrile |
US8088943B2 (en) | 2008-01-15 | 2012-01-03 | Invista North America S.A R.L. | Hydrocyanation of pentenenitriles |
US8247621B2 (en) | 2008-10-14 | 2012-08-21 | Invista North America S.A.R.L. | Process for making 2-secondary-alkyl-4,5-di-(normal-alkyl)phenols |
US8338636B2 (en) | 2009-08-07 | 2012-12-25 | Invista North America S.A R.L. | Hydrogenation and esterification to form diesters |
US8913501B2 (en) | 2012-08-16 | 2014-12-16 | International Business Machines Corporation | Efficient urgency-aware rate control scheme for multiple bounded flows |
Also Published As
Publication number | Publication date |
---|---|
CN1852242A (zh) | 2006-10-25 |
EP1814265B1 (en) | 2011-03-02 |
WO2007051374A8 (fr) | 2007-07-19 |
US7684348B2 (en) | 2010-03-23 |
EP1814265A1 (en) | 2007-08-01 |
CN101156382A (zh) | 2008-04-02 |
US20090010161A1 (en) | 2009-01-08 |
ATE500667T1 (de) | 2011-03-15 |
DE602006020400D1 (de) | 2011-04-14 |
CN100558069C (zh) | 2009-11-04 |
CN100433718C (zh) | 2008-11-12 |
EP1814265A4 (en) | 2008-09-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2007051374A1 (en) | A method for guaranteeing classification of service of the packet traffic and the method of rate restriction | |
US6970424B2 (en) | Method and apparatus to minimize congestion in a packet switched network | |
KR100644445B1 (ko) | 다-임계값 리키 버킷을 사용하는 클래스-기초 속도 제어 | |
US6714517B1 (en) | Method and apparatus for interconnection of packet switches with guaranteed bandwidth | |
US6654374B1 (en) | Method and apparatus to reduce Jitter in packet switched networks | |
US7142563B1 (en) | Service interface for QoS-driven HPNA networks | |
US20020059408A1 (en) | Dynamic traffic management on a shared medium | |
US20030185249A1 (en) | Flow control and quality of service provision for frame relay protocols | |
US20060013133A1 (en) | Packet-aware time division multiplexing switch with dynamically configurable switching fabric connections | |
WO2002098080A1 (en) | System and method for scheduling traffic for different classes of service | |
US20100182904A1 (en) | Logical transport resource traffic management | |
JP4899649B2 (ja) | リジリエントパケットリングトラフィック管理のための装置及び方法 | |
US8228797B1 (en) | System and method for providing optimum bandwidth utilization | |
CN101026558A (zh) | 一种实现海量端口反压的方法及其装置 | |
US6947380B1 (en) | Guaranteed bandwidth mechanism for a terabit multiservice switch | |
JP4208710B2 (ja) | フレーム伝送装置 | |
CN115695201A (zh) | 一种基于FlexE Channel优先级的带宽资源分配方法和系统 | |
WO2022135202A1 (zh) | 业务流的调度方法、装置及系统 | |
US7630303B2 (en) | Communication control method and network system | |
WO2006042470A1 (fr) | Procede de transmission equitable de flux dans un reseau local en boucle mpls | |
WO2006042472A1 (fr) | Procede de mise en oeuvre de transmission satisfaisante de flux dans le reseau en anneau mpls | |
CN117596208A (zh) | 一种适用于确定性网络的无损流量控制方法 | |
CN116260769A (zh) | 面向分布式网络的时间确定性流量突发整形方法 | |
Jiang et al. | Spatial reuse and fairness in resilient packet ring | |
Yu et al. | Spatial shaping with dynamic priority VDQ in RPR |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2006741900 Country of ref document: EP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWP | Wipo information: published in national office |
Ref document number: 2006741900 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 200680011899.0 Country of ref document: CN |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 11568311 Country of ref document: US |