WO2008000127A1

WO2008000127A1 - An apparatus and method for dispatching message

Info

Publication number: WO2008000127A1
Application number: PCT/CN2007/001171
Authority: WO
Inventors: Chuntao Wang; Chao Hou; Jianbing Wang
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2006-06-19
Filing date: 2007-04-11
Publication date: 2008-01-03
Also published as: CN1968186A

Abstract

An apparatus for dispatching message includes Ln dispatchers for dispatching and aggregating to transmit message under stratification and Ln-1 dispatchers for dispatching and being aggregated to transmit message under stratification. Each of the said Ln dispatcher aggregates mulitple Ln-1 dispatchers to dispatch the message to be transmitted. Each of the Ln-1 dispatchers for aggregating to transmit message is corresponding with a group of polymerized queues, and the different queues in the group store the message to be transimtted respectively, and the message is dispatched by the said Ln-1 dispatchers that are aggregated to transmit message. Further, a corresponding method for dispatching message is also disclosed. The present invention may be applied in the situation of using the aggregation technique, and it can realize dispatching the message to be transmitted in terms of the stratification, and guarantee the QOS of the service of applying the aggregation technique.

Description

Device and method for scheduling messages

This application claims priority to Chinese Patent Application No. 200610036002.1, entitled "A Device and Method for Dispatching Dispatched Messages", filed on June 19, 2006, the entire contents of which are incorporated by reference. In this application.

Technical field

The present invention relates to a message scheduling technique, and more particularly to an apparatus and method for scheduling messages.

Background technique

In the complex environment of Internet packet switching, network congestion is more common. Congestion makes traffic unable to obtain resources in time, which is the source of service performance degradation. Congestion may cause the following negative effects: Congestion increases packet transmission delay and delay jitter, and excessive delay causes packet retransmission; congestion The effective throughput of the network is reduced, causing damage to network resources; increased congestion consumes a large amount of network resources (especially storage resources), and unreasonable resource allocation may even cause the system to fall into a resource deadlock and collapse. However, in the complex environment where packet switching and multi-user services coexist, congestion is common. When the network is congested, it must be managed and controlled. Common solutions usually use queuing technology.

At present, according to the topology structure of a typical broadband access network, the prior art defines a multi-layer "many-to-one" tree queue scheduling architecture, so that digital subscriber line (DSL) broadband access technology can be As shown in FIG. 1 , the tree queue scheduling architecture can truly reflect the forwarding model of a typical broadband service, so as to meet the QOS requirements of different services.

Specifically, the scheduling process of the hierarchical queue scheduling in the prior art is as follows:

1) First, the device schedules a physical port scheduler (scheduler) 101 according to the rate of the physical port. Once the physical port scheduler is scheduled, the next-level scheduler owned by the physical port scheduler is scheduled according to the configured scheduling algorithm— - VP scheduler 102; the above scheduling algorithm, for example: Weighted Fair Queuing (WFQ), each virtual path (VP) scheduler is configured with different weights;

2) When one of the VP schedulers 102 is scheduled, the next-level scheduler owned by the VP scheduler, the virtual circuit (VC) group, is scheduled according to the configured scheduling algorithm. Levelter 103;

3) When one of the VC group schedulers 103 is scheduled, the next scheduled scheduler owned by the VC group scheduler, the VC scheduler 104, is scheduled according to the configured scheduling algorithm;

4) When one of the VC schedulers 104 is scheduled, the next-stage scheduler owned by the VC scheduler, the Session scheduler 105, is scheduled according to the configured scheduling algorithm;

5) Finally, when one of the Session Schedulers 105 is scheduled, the queue owned by the Session Scheduler 105, the flow classification queue, will be scheduled according to the configured scheduling algorithm, and a message will be dispatched from the flow classification queue. , used to send on a physical port.

Although the prior art defines a hierarchical queue scheduling model, in actual application, the above scheduling process does not need to be strictly adhered to, because the hierarchical level of hierarchical scheduling is not fixed, and needs to be determined by the topology level of the network, and each layer scheduling The role of the device also needs to be mapped according to the network structure.

On the other hand, when the bandwidth required by the link exceeds the bandwidth that can be provided by a single physical interface or channel, the prior art may use aggregation technology to bundle one or more physical interfaces or channels to form a logical interface or channel. This logical interface or channel can provide higher bandwidth for the link, and the available bandwidth of the link is the sum of the bandwidth of all physical interfaces or channels. Currently, the main aggregation technologies are: Ethernet Trunk, Point-to-Point Link Multilinl PPP, Multilinlc Frame Relay, etc. Because in the case of using aggregation technology, different packets of the same VP are transmitted through different physical interfaces. Between the VP and the physical interface is a "one-to-many, or "many-to-many" mapping relationship instead of "multiple". For a "transfer model that correctly reflects the business using aggregation technology, it cannot guarantee the business of using aggregation technology.

QOS ₀

Summary of the invention

An embodiment of the present invention provides an apparatus and a method for scheduling a message, so that, in the case of using the aggregation technology, the packet to be sent can be hierarchically scheduled to ensure the Q0S of the service using the aggregation technology.

An apparatus for scheduling a message according to an embodiment of the present invention includes: an Ln scheduler according to hierarchical scheduling and an Ln-1 scheduler according to hierarchical scheduling, where each Ln scheduler aggregates multiple Ln-1 schedulers Scheduling packets to be sent, where: Each Ln scheduler corresponds to a group of aggregation queues, and each aggregation queue in the group corresponds to one Ln-1 scheduler, and the packets to be sent scheduled by the Ln-1 scheduler are saved.

Correspondingly, a method for scheduling a message according to an embodiment of the present invention includes:

According to the hierarchical scheduling message transmission, when the round-robin scheduling is performed by the Ln scheduler, the Ln scheduler dispatches the to-be-sent packet into an aggregation queue of the corresponding group of aggregation queues; wherein each aggregation queue in the group corresponds to An Ln-1 scheduler;

When the round robin is scheduled to the Ln-1 scheduler, the packets to be sent are scheduled from the corresponding aggregate queue. In the embodiment of the present invention, the Ln scheduler of each aggregated transmission packet corresponds to a group of aggregation queues, and the different aggregation queues in the group respectively store the packets to be sent scheduled by the Ln-1 scheduler of the corresponding aggregated transmission message. Since the aggregation queue and the Ln-1 scheduler perform packet scheduling according to the "many-to-one" correspondence, the hierarchical queue scheduling architecture can be extended, and the hierarchical queue scheduling architecture can be applied to the occasion of using link aggregation technology. It can realize the forwarding model of the service that correctly reflects the aggregation technology, and guarantee the QOS of the service using the aggregation technology.

DRAWINGS

1 is a schematic diagram of hierarchical scheduling in the prior art;

2 is a schematic diagram of an application scenario using an aggregation technology according to an embodiment of the present invention;

3 is a schematic diagram of scheduling, by using an aggregation queue, a message to be sent according to an embodiment of the present invention;

4 is a schematic diagram of a specific network topology structure for scheduling a message to be sent in an aggregation situation according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of hierarchical scheduling of link aggregation in the network environment shown in FIG. 4.

detailed description

The embodiment of the present invention enables the Ln scheduling of each aggregated transmission message by extending the hierarchical scheduling architecture.

The Ln-1 scheduler schedules the packets to be sent. Because the aggregation queue and the Ln-1 scheduler can perform message scheduling according to the "many-to-one" hierarchical scheduling correspondence, the hierarchical scheduling architecture can be applied to When the link aggregation technology is used, and the forwarding model of the service is correctly reflected, the QOS of the service is guaranteed, which is described in detail below.

When the link aggregation technology is used, different packets of the same link are transmitted through different physical interfaces or channels of the lower layer. To ensure the QOS of the service, you need to be able to aggregate the queues in a hierarchical manner. This "one-to-many" or "many-to-many" mapping relationship is reflected in the scheduling architecture. Referring to FIG. 2, FIG. 2 is a schematic diagram of an application scenario using an aggregation technique, and FIG. 2 illustrates a "many-to-many" mapping relationship between two adjacent layers in a hierarchical scheduling architecture. That is, there is a "many-to-many" mapping relationship between the Ln scheduler 220 and the Ln-1 scheduler 210.

In the hierarchical scheduling, both the Ln scheduler 220 and the Ln-1 scheduler 210 need to perform scheduling according to a specified rate. In the case of a "many-to-one" mapping between the Ln scheduler and the Ln-1 scheduler, It is implemented according to the scheduling method defined by the prior art. However, in the case where there is a "one-to-many" or "many-to-many" mapping between the Ln tuner 220 and the Ln-1 scheduler 210, that is, one Ln scheduler aggregates a plurality of Ln-1 schedulers, Queue scheduling cannot be performed directly according to the scheduling method defined by the prior art.

To solve this problem, referring to FIG. 3, an embodiment of the present invention introduces an aggregation queue (Aggregation Queue) 200 between the Ln scheduler 220 that aggregates the transmitted message and the Ln-1 scheduler 210 that aggregates the transmitted message, that is, each The Ln scheduler 220 that aggregates the transmitted packets corresponds to a set of aggregated queues and packets sent within the group.

For each group of aggregate queues corresponding to each Ln scheduler, the number of aggregate queues included in each group of aggregate queues is determined by the number of Ln-1 schedulers in the aggregation group, and different aggregation queues in the group correspond to different Ln-1 schedules. In the specific implementation, the aggregation queue and the Ln-1 scheduler in the aggregation queue group of the Ln scheduler may correspond to 1:1 or n:1, but no matter which correspondence, each group in the group The aggregation queue corresponds to an Ln-1 scheduler, and is not described here.

The following describes the method for scheduling a packet in the embodiment of the present invention. In this embodiment, the Ln scheduler aggregates and sends a packet, and the Ln-1 scheduler is aggregated to send a packet. The specific scheduling packet sending process is as follows:

1) Start hierarchical queue scheduling, and when the round robin schedules the Ln scheduler to send packets, the Ln scheduler schedules according to the specified rate;

2) After the Ln scheduler successfully dispatches a packet, the Ln-1 scheduler identifier (or its aggregate queue identifier) is put into the corresponding aggregation queue according to the packet;

3) When the hierarchical queue scheduling is rotated to the Ln-1 scheduler that is configured to send the packet, the Ln-1 scheduler schedules the packet to be sent according to the specified aggregation queue, and the present invention implements the packet to be sent. In the example, the correspondence between the aggregation queue and the Ln-1 scheduler is "many-to-one". Therefore, the packets can be scheduled according to the hierarchical queue scheduling mode, and are not described here. The "many-to-many" mapping is described above. The "one-to-many" mapping is a special case of the "many-to-many" mapping. According to the above process, those skilled in the art can easily implement it, and details are not described herein.

The following is a practical application scenario to illustrate the hierarchical queue scheduling process in the link aggregation scenario. Figure 4 is a schematic diagram of a network topology structure using hierarchical aggregation of link aggregation, as shown in Figure 4. It can be seen that Gigabit Ethernet (GE) Trunk technology is used between the Broadband Remote Access Server (BRAS) 401 and the User Access Layer Device (LAN) 402. And implement hierarchical scheduling on the BRAS device.

By analyzing the topological structure of the above network, a hierarchical scheduling model supporting link aggregation can be designed. As shown in Figure 5, each level of the scheduler corresponds to the outgoing port of the hierarchical device, and the BRAS 410 is ready to synchronize the slave. Trunk processing and traffic classification processing, trunk processing, and traffic classification are performed before the packets received on the Packet Over Synchronous optical network (Sync Hourous Domain) (POS) interface are forwarded to the GE interface. The process determines the GE aggregation queue ID and the traffic classification queue ID. After the traffic classification process is complete, the packet and the GE aggregation queue ID are sent to the traffic classification queue.

The processing procedure of the BRAS for hierarchical queue scheduling in this embodiment is as follows:

1) First: BRAS 410 schedules each Lanswitch hierarchical scheduler 402 according to the configured rate (eg: 100Mbps);

2) Once one of the Lanswtich hierarchical schedulers 402 is scheduled, each digital subscriber line access multiplexing subordinate to the Lanswitch hierarchical scheduler 402 will be scheduled according to a configured algorithm (e.g., a Round Robin scheduling algorithm). (Digital Subscriber Line Access Multiplexer, DSLAM) hierarchical scheduler 403;

3) When one of the DSLAM hierarchical schedulers 403 is scheduled, the configured algorithm, such as the WFQ scheduling algorithm, each Customer Premises Equipment (CPE) hierarchical scheduler 404 is configured with different weights. Scheduling each CPE hierarchical scheduler 404 subordinate to the DSLAM hierarchical scheduler 403;

4) When one of the CPE hierarchical schedulers 404 is scheduled, a packet will be fetched from the owned traffic classification queue according to the configured algorithm (for example, an absolute priority scheduling algorithm), and the GE aggregation queue according to the packet ID is entered into the corresponding GE aggregation queue 400; 5) While performing 1) ~ 4) scheduling, the BRAS schedules each BRAS hierarchical scheduler 401 according to the configured rate (eg: 1000 Mbps), and once one of the BRAS hierarchical schedulers 401 is scheduled, will be configured according to The algorithm (for example, the Round Robin scheduling algorithm) takes a packet from the owned GE aggregation queue 400 and sends it on the corresponding physical port, so that a complete hierarchical queue scheduling process supporting link aggregation is completed. It is.

In the embodiment of the present invention, the Ln scheduler of each aggregated transmission packet corresponds to a group of aggregation queues, and the different queues in the group respectively save the corresponding Ln-1 scheduler of the aggregated transmission packet to be scheduled to be sent, due to aggregation. The queue and the Ln-1 scheduler perform packet scheduling according to the "many-to-one" correspondence, so that the hierarchical queue scheduling architecture can be extended, and the hierarchical queue scheduling architecture can be applied to the occasion of using link aggregation technology. A forwarding model that correctly reflects the business using the aggregation technology and guarantees the QOS of the service using the aggregation technology.

The above is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or within the technical scope disclosed by the present invention. Alternatives are intended to be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

Rights request

A device for scheduling messages, comprising: an Ln scheduler according to hierarchical scheduling and an Ln-1 scheduler according to hierarchical scheduling, wherein each Ln scheduler aggregates multiple Ln-1 schedulers to be scheduled to be sent. Message, which is characterized by

Each Ln scheduler corresponds to a group of aggregation queues; each aggregation queue in the group corresponds to an Ln-1 scheduler, and the packets to be sent scheduled by the Ln-1 scheduler are saved.

.

2. The apparatus for scheduling messages according to claim 1, wherein the number of aggregation queues in a group of aggregation queues corresponding to each Ln scheduler is determined by the number of Ln-1 schedulers.

The device for scheduling messages according to claim 2, wherein the number of aggregate queues in the group of aggregate queues corresponding to each Ln scheduler is corresponding to the number of Ln-1 schedulers according to 1:1. .

The apparatus for scheduling messages according to claim 2, wherein the number of aggregate queues in a group of aggregate queues corresponding to each Ln scheduler and the number of Ln-1 schedulers are in accordance with n: 1 .

The device for scheduling a message according to claim 1, wherein the packet to be sent is a corresponding Ln scheduler according to an Ln-1 scheduler identifier or an aggregate queue identifier of a packet to be sent. Aggregate the queue.

The apparatus for scheduling messages according to any one of claims 1-5, wherein the Ln scheduler is a virtual channel scheduler, and the Ln-1 scheduler is a physical port scheduler.

A method for scheduling a message, comprising:

The Ln scheduler schedules the to-be-sent packet into an aggregation queue of the corresponding group of aggregation queues, where each aggregation in the group is scheduled according to the hierarchical scheduling of the packets to be sent. The queue corresponds to an Ln-1 scheduler;

When the round-robin scheduling is performed to the Ln-1 scheduler, the message to be sent is scheduled from the corresponding aggregate queue.

The method for scheduling messages according to claim 7, wherein the number of aggregate queues in a group of aggregate queues corresponding to each Ln scheduler is determined by the number of Ln-1 schedulers.

The method for scheduling a message according to claim 8, wherein the number of the aggregated queues in the group of aggregated queues corresponding to each Ln scheduler is corresponding to the number of Ln-1 schedulers. The system is 1:1.

The method for scheduling a message according to claim 8, wherein the correspondence between the number of aggregate queues in the group of aggregate queues corresponding to each Ln scheduler and the number of Ln-1 schedulers is n. : 1.

The method for sending a scheduling message according to claim 7, wherein the Ln scheduler schedules a packet to be sent into an aggregation queue of a corresponding group of aggregation queues according to a packet to be sent. Ln-1 scheduler ID or aggregate queue ID.

The method for scheduling messages according to any one of claims 7 to 11, wherein the Ln scheduler is a virtual channel scheduler, and the Ln-1 scheduler is a physical port scheduler.

The method for scheduling messages according to claim 7, wherein the Ln scheduler and the Ln-1 scheduler both perform scheduling at a specified rate.