CN103166875A - Double-layer isolation method for network resources based on net field programmable gate array - Google Patents
Double-layer isolation method for network resources based on net field programmable gate array Download PDFInfo
- Publication number
- CN103166875A CN103166875A CN2013100778761A CN201310077876A CN103166875A CN 103166875 A CN103166875 A CN 103166875A CN 2013100778761 A CN2013100778761 A CN 2013100778761A CN 201310077876 A CN201310077876 A CN 201310077876A CN 103166875 A CN103166875 A CN 103166875A
- Authority
- CN
- China
- Prior art keywords
- packet
- stream
- conflux
- formation
- bandwidth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
本发明公开了一种基于可编程交换机的网络资源的双层隔离方法，该方法将网络资源分为聚流层和流层，在聚流层调度时为每一个聚流预留指定带宽，在流层调度时让聚流内的多个流根据其权重竞争带宽。本发明使用低成本的NetFPGA硬件设备，实现了网络资源的灵活、公平且可控的分配，降低了网络资源分配的实现成本，其实现方法简便，手段灵活，网络服务质量能得到显著保证，且与网络层协议无关。The invention discloses a double-layer isolation method of network resources based on a programmable switch. The method divides network resources into a flow aggregation layer and a flow layer, and reserves a specified bandwidth for each flow aggregation when scheduling the flow aggregation layer. When the stream layer is scheduled, multiple streams in the aggregated stream compete for bandwidth according to their weights. The present invention uses low-cost NetFPGA hardware equipment to realize flexible, fair and controllable allocation of network resources, and reduces the implementation cost of network resource allocation. The implementation method is simple and flexible, and the network service quality can be significantly guaranteed, and It has nothing to do with the network layer protocol.
Description
Technical field
The present invention relates to technical field of the computer network, relate in particular to a kind of double-deck partition method of the Internet resources based on programmable switch (Net Field Programmable Gate Array, NetFPGA).
Background technology
The Internet has been obtained huge success in the past few decades, the platform that it provides a brand-new exchange message, received information for people.Extensively popularizing of it promoted development and the application of network technology.Yet huge network has also limited its development in the future simultaneously.Such as, with scale and the size of current network, for each Network guarantees service quality very difficult.Traditional QoS implementation method is generally based on the model of Integration Services or Differentiated Services.Integration Services keeps independent resource for each business, but easily causes the wasting of resources; Differentiated Services is according to label flexible allocation resources such as priority, but the Internet resources that isolate for single business reservation.
The model of Integration Services and Differentiated Services respectively has superiority, but lacks flexibility, can't adapt to changeable network demand; On the other hand, the distribution method of Internet resources based on software simulation or theoretical the derivation, lacks concrete implementation method mostly at present.
Summary of the invention
The object of the invention is to for the deficiencies in the prior art, a kind of double-deck partition method of the Internet resources based on programmable switch is provided.
The objective of the invention is to be achieved through the following technical solutions: a kind of double-deck partition method of the Internet resources based on programmable switch comprises the following steps:
Step 1: packet enters programmable switch by one of 8 input ports: programmable switch has 8 input ports, comprises 4 1G ethernet ports and 4 DMA ports; Switch adopts polling mode to select packet to enter network layer from one of 8 input ports in turn;
Step 2: network layer protocol is selected the output port of packet according to routing table according to the packet head, and in output queue corresponding to packet head mark;
Step 3: use double-deck Internet resources partition method to mark the packet of output queue dispatch; This step realizes by following substep:
(3.1) obtain the stream number of packet: if be not with the VLAN label from the packet of network layer output, think that the stream number of this packet is 0; If the data band VLAN label from network layer output, use low its stream number of 2 bit-identifies of the VID field of VLAN: be not or not 00 packet is identified as flowing 0 with VLAN label or low 2 of VID, 2, VID end is that 01 packet is identified as flowing 1,2, VID end is that to be identified as flowing 2,2, VID end be that 03 packet is identified as flowing 3 for 02 packet;
(3.2) set up a scheduling queue for each stream, packet is sent into corresponding formation according to stream number;
(3.3) convection current is divided into groups, and one or more streams are combined into a conflux;
(3.4) use weighting round-robin mode to distribute the bandwidth of each conflux: establish the timeslice that a next round turns and add up to N, the nominated bandwidth of any one conflux i is b
i, total bandwidth is b, turning middle distribution timeslice in a next round for each conflux is b
i/ b * N; When taking turns to a conflux i acceptance service, all streams that its time will be belonged to conflux i use; When timeslice used, conflux i stopped the acceptance service, switched to next conflux and accepted service;
(3.5) within the service time that conflux i is assigned to, belong to the service time of each stream f according to the weight allocation of stream: the interior a plurality of streams of conflux are queued up and are accepted poll, for each stream arranges a balance counter; When a stream begins to receive service, according to the weight of this stream, add a difference (difference is directly proportional to the weight size) to the balance counter of this stream; For a stream that is being sent out, packet of every transmission, balance counter will reduce the byte number of this packet, when the difference counter be finished or the formation of this stream when empty, just switch to next stream; The increase of stream and minimizing can make the bandwidth that each stream obtains change, and make like this bandwidth in conflux to be taken full advantage of by the inner stream that exists;
(3.6) packet in certain formation is full of formation, and when having new data packets to enter this formation, this formation will abandon the oldest packet in this formation;
Step 4: the packet through scheduling enters output queue: programmable switch has 8 output ports, comprises 4 1G ethernet ports and 4 DMA ports; According to the output port mark that adds in step 2, send into specified queue.
The invention has the beneficial effects as follows, the present invention uses NetFPGA hardware device cheaply, realized flexible, the fair and controlled distribution of Internet resources, reduced the cost of realizing of Resource Allocation in Networks, its implementation is easy, means are flexible, and network service quality can significantly be guaranteed, and have nothing to do with network layer protocol.
Description of drawings
Fig. 1 is based on the double-deck partition method flow chart of the Internet resources of NetFPGA hardware.
Embodiment
The below describes the present invention with reference to the accompanying drawings in detail.
The present invention is based on the double-deck partition method of the Internet resources of programmable switch, comprise the following steps:
Step 1: packet enters programmable switch by one of 8 input ports.
Programmable switch has 8 input ports, comprises 4 1G ethernet ports and 4 DMA ports.Switch adopts polling mode to select packet to enter network layer from one of 8 input ports in turn.
Step 2: network layer protocol is selected the output port of packet according to routing table according to the packet head, and in output queue corresponding to packet head mark (for step 4).
Step 3: use double-deck Internet resources partition method to mark the packet of output queue dispatch.
This step is core of the present invention, is divided into following substep.
1) obtain the stream number of packet.
Each is from the packet of network layer output or not with VLAN label (think that stream number is 0 this moment) or with VLAN label (using low its stream number of 2 bit-identifies of the VID field of VLAN).Be not or not to be identified as flowing 2,0, VID end be that to be identified as flowing 2,1, VID end be that to be identified as flowing 2,2, VID end be that 03 packet is identified as flowing 3 for 02 packet for 01 packet for 00 packet with low 2 of VLAN label or VID.
2) set up a scheduling queue for each stream, packet is sent into corresponding formation according to stream number.
Formation can directly use the Core Generator software of match company of SEL (Xilinx) to utilize block storage (BRAM) resource on NetFPGA automatically to generate fifo queue on programmable switch.Packet is sent into corresponding formation.
3) convection current is divided into groups, and one or more streams are combined into a conflux.
4) use weighting round-robin mode to distribute the bandwidth of each conflux.
If the timeslice that a next round turns adds up to N, the nominated bandwidth of any one conflux i is b
i, total bandwidth is b, turning middle distribution timeslice in a next round for each conflux is b
i/ b * N.
When taking turns to a conflux i acceptance service, all streams that its time will be belonged to conflux i use; When timeslice used, conflux i stopped the acceptance service, switched to next conflux and accepted service.By this method, each conflux is accepted the time of service and can clearly be specified, and like this, the bandwidth of single conflux just can be guaranteed.
5) within the service time that conflux i is assigned to, belong to the service time of each stream f according to the weight allocation of stream.Here use classical difference polling method.The interior a plurality of streams of conflux are queued up and are accepted poll, for each stream arranges a balance counter (Deficit Counter).When a stream begins to receive service, according to the weight of this stream, add a difference (difference is directly proportional to the weight size) to the balance counter of this stream.For a stream that is being sent out, packet of every transmission, balance counter will reduce the byte number of this packet, when the difference counter be finished or the formation of this stream when empty, just switch to next stream.The increase of stream and minimizing can make the bandwidth that each stream obtains change, and make like this bandwidth in conflux to be taken full advantage of by the inner stream that exists.
Double-deck structure makes this method not only can adapt to for example requirement of the strict allocated bandwidth of virtual net (by distributing a conflux to each virtual net), also can adapt to the demand according to the weight allocation bandwidth simultaneously.
6) packet in certain formation is full of formation, and when having new data packets to enter this formation, this formation will abandon the oldest packet in this formation.
Step 4: the packet through scheduling enters output queue.
Programmable switch has 8 output ports, comprises 4 1G ethernet ports and 4 DMA ports.According to the output port mark that adds in step 2, send into specified queue.
Claims (1)
1. the double-deck partition method based on the Internet resources of programmable switch, is characterized in that, comprises the following steps:
Step 1: packet enters programmable switch by one of 8 input ports: programmable switch has 8 input ports, comprises 4 1G ethernet ports and 4 DMA ports; Switch adopts polling mode to select packet to enter network layer from one of 8 input ports in turn;
Step 2: network layer protocol is selected the output port of packet according to routing table according to the packet head, and in output queue corresponding to packet head mark;
Step 3: use double-deck Internet resources partition method to mark the packet of output queue dispatch; This step realizes by following substep:
(3.1) obtain the stream number of packet: if be not with the VLAN label from the packet of network layer output, think that the stream number of this packet is 0; If the data band VLAN label from network layer output, use low its stream number of 2 bit-identifies of the VID field of VLAN: be not or not 00 packet is identified as flowing 0 with VLAN label or low 2 of VID, 2, VID end is that 01 packet is identified as flowing 1,2, VID end is that to be identified as flowing 2,2, VID end be that 03 packet is identified as flowing 3 for 02 packet;
(3.2) set up a scheduling queue for each stream, packet is sent into corresponding formation according to stream number;
(3.3) convection current is divided into groups, and one or more streams are combined into a conflux;
(3.4) use weighting round-robin mode to distribute the bandwidth of each conflux: establish the timeslice that a next round turns and add up to N, the nominated bandwidth of any one conflux i is b
i, total bandwidth is b, turning middle distribution timeslice in a next round for each conflux is b
i/ b * N; When taking turns to a conflux i acceptance service, all streams that its time will be belonged to conflux i use; When timeslice used, conflux i stopped the acceptance service, switched to next conflux and accepted service;
(3.5) within the service time that conflux i is assigned to, belong to the service time of each stream f according to the weight allocation of stream: the interior a plurality of streams of conflux are queued up and are accepted poll, for each stream arranges a balance counter; When a stream begins to receive service, according to the weight of this stream, add a difference (difference is directly proportional to the weight size) to the balance counter of this stream; For a stream that is being sent out, packet of every transmission, balance counter will reduce the byte number of this packet, when the difference counter be finished or the formation of this stream when empty, just switch to next stream; The increase of stream and minimizing can make the bandwidth that each stream obtains change, and make like this bandwidth in conflux to be taken full advantage of by the inner stream that exists;
(3.6) packet in certain formation is full of formation, and when having new data packets to enter this formation, this formation will abandon the oldest packet in this formation;
Step 4: the packet through scheduling enters output queue: programmable switch has 8 output ports, comprises 4 1G ethernet ports and 4 DMA ports; According to the output port mark that adds in step 2, send into specified queue.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310077876.1A CN103166875B (en) | 2013-03-12 | 2013-03-12 | A kind of double-layer isolation method of the Internet resources based on programmable switch |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310077876.1A CN103166875B (en) | 2013-03-12 | 2013-03-12 | A kind of double-layer isolation method of the Internet resources based on programmable switch |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103166875A true CN103166875A (en) | 2013-06-19 |
| CN103166875B CN103166875B (en) | 2015-10-28 |
Family
ID=48589625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310077876.1A Active CN103166875B (en) | 2013-03-12 | 2013-03-12 | A kind of double-layer isolation method of the Internet resources based on programmable switch |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103166875B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115484157A (en) * | 2022-09-14 | 2022-12-16 | 浙江大学 | Sketch general configuration method based on programmable switch |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101102275A (en) * | 2007-06-25 | 2008-01-09 | 中兴通讯股份有限公司 | Method for Realizing Multilevel Scheduling on Ethernet Switch Chip |
| CN101938403A (en) * | 2009-06-30 | 2011-01-05 | 中国电信股份有限公司 | Assurance method of multi-user and multi-service quality of service and service access control point |
| CN102546423A (en) * | 2012-01-20 | 2012-07-04 | 福建星网锐捷网络有限公司 | Method and device for queue scheduling and network device |
-
2013
- 2013-03-12 CN CN201310077876.1A patent/CN103166875B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101102275A (en) * | 2007-06-25 | 2008-01-09 | 中兴通讯股份有限公司 | Method for Realizing Multilevel Scheduling on Ethernet Switch Chip |
| CN101938403A (en) * | 2009-06-30 | 2011-01-05 | 中国电信股份有限公司 | Assurance method of multi-user and multi-service quality of service and service access control point |
| CN102546423A (en) * | 2012-01-20 | 2012-07-04 | 福建星网锐捷网络有限公司 | Method and device for queue scheduling and network device |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115484157A (en) * | 2022-09-14 | 2022-12-16 | 浙江大学 | Sketch general configuration method based on programmable switch |
| CN115484157B (en) * | 2022-09-14 | 2023-06-02 | 浙江大学 | General configuration method of sketch based on programmable switch |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103166875B (en) | 2015-10-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2302843B1 (en) | Method and device for packet scheduling | |
| CN104378309B (en) | Method, system and related equipment for realizing QoS in OpenFlow network | |
| CN101009655B (en) | Traffic scheduling method and device | |
| CN106533982B (en) | The dynamic queue's dispatching device and method borrowed based on bandwidth | |
| CN107733689A (en) | Dynamic weighting polling dispatching strategy process based on priority | |
| US9940471B2 (en) | Virtual output queue authorization management method and device, and computer storage medium | |
| CN104753809B (en) | The method and device of token is added in a kind of traffic shaping | |
| CN105490962B (en) | A kind of QoS management methods based on OpenFlow networks | |
| CN107483363A (en) | A hierarchical weighted round robin scheduling device and method | |
| CN102611605A (en) | Scheduling method, device and system of data exchange network | |
| CN101714947A (en) | Extensible full-flow priority dispatching method | |
| CN102546395A (en) | Service scheduling method and service scheduling device based on Layer 2 Virtual Private Networks (L2VPN) | |
| CN107204930B (en) | Token adding method and device | |
| CN109391559A (en) | The network equipment | |
| CN102769566A (en) | Method and device for configuring multilevel scheduling system, and method and device for changing configuration of multilevel scheduling system | |
| US9128755B2 (en) | Method and apparatus for scheduling resources in system architecture | |
| CN103166875A (en) | Double-layer isolation method for network resources based on net field programmable gate array | |
| CN101557346A (en) | Polling-based packet queue output scheduling method and packet switching node | |
| Yang et al. | A new class of priority-based weighted fair scheduling algorithm | |
| Luangsomboon et al. | HLS: a packet scheduler for hierarchical fairness | |
| CN107483361A (en) | A kind of scheduling model construction method and device | |
| Jha et al. | New Queuing Technique for Improving Computer Networks QoS | |
| CN103326962B (en) | Diversified service switching method | |
| CN102594670A (en) | Multiport multi-flow scheduling method, device and equipment | |
| CN104410584A (en) | QoS queue level scheduling method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |