CN103546337A - Flow demand estimation method for packet and circuit hybrid switching system - Google Patents
Flow demand estimation method for packet and circuit hybrid switching system Download PDFInfo
- Publication number
- CN103546337A CN103546337A CN201310445946.4A CN201310445946A CN103546337A CN 103546337 A CN103546337 A CN 103546337A CN 201310445946 A CN201310445946 A CN 201310445946A CN 103546337 A CN103546337 A CN 103546337A
- Authority
- CN
- China
- Prior art keywords
- voq
- value
- traffic demand
- switching system
- length
- 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.)
- Pending
Links
Images
Abstract
The invention provides a flow demand estimation method for a packet and circuit hybrid switching system. The method includes the information sampling process, namely reading and recording the length of a virtual outputting queue periodically; the information extracting and estimating process, namely extracting the growth rate of the virtual outputting queue and estimating the flow rate of the queue; a flow demand estimating process, namely estimating total flow demand according to the results of the sampling, extracting and estimating, and extracting a stable portion of the total flow demand estimating value to use for circuit distribution. According to the method, caching and flowing rate which influence the hybrid switching system performances are considered, and the method has broad applicability; all scheduling results, namely resource distributing information, are fed back to be used for participating flow demand estimation, a feedback system is formed, and the results are more accurate; meanwhile, information in an existing packet and circuit hybrid switching system is utilized to perform flow demand estimation, additional support by hardware is omitted, and the method is easy to realize accordingly.
Description
Technical field
The present invention relates to the method for communication technical field, especially, in switch system, be specifically related to the traffic demand method of estimation in a kind of grouping and circuit hybrid switching system.
Background technology
At communication technical field, than pure packet switching system or pure circuit switching system, grouping and circuit hybrid switching system have a lot of advantages, as carried out statistic multiplexing, improve line efficiency, connection-oriented service guarantee service quality (QoS) is provided simultaneously, thereby there are less time delay, higher throughput, lower backlog rate, better QoS assurance, therefore obtained universal rapidly.Along with network technology more intellectuality and high speed, in high-speed mixing switching system, packet switching and Circuit-switched intelligence switch to one of key technology for technology evolution, and more intelligent switching depends on more efficient traffic demand method of estimation.
Traffic demand is estimated the quality as the direct decision systems performance of important foundation of hybrid switching system, especially in hybrid switching system at a high speed.In high-speed mixing switching system, because the arrival interval of data is extremely short, simultaneous streaming huge amount, carrying out traffic demand accurately estimates, its very expensive, particularly, in data center network, ten hundreds of data flow is carried out to traffic demand accurately and estimate, its network cost, overhead, energy consumption are all abnormal huge.In this case, propose a kind ofly can be applied to efficiently high-speed packet and circuit hybrid switching system, the traffic demand method of estimation that can tackle diverse network traffic characteristic is simultaneously very necessary, is also very significant.
Through existing literature search is found, G.Wang, the people such as D.G.Andersen are in the < < ACM SIGCOMM Computer Communication Review-SIGCOMM'10(ACM of Association for Computing Machinery data communication special interest group compunication review-data communication special interest group 10) delivered the article that is entitled as " c-Through:part-time optics in data centers " on > > the 40th volume the 4th phase 327-338 page, the traffic demand method of estimation taking based on buffer memory has been proposed in article, by detecting the buffer memory of each VOQ (VOQ) in current high-speed mixing switching system, take the factor that situation is used as traffic demand estimation, the advantage of the method is simple, easy to implement, but when flow is highly stable, buffer memory takies and can go to zero, the method can lose efficacy.
Find by literature search again, N.Farrington, the people such as G.Porter are in the < < ACM SIGCOMM Computer Communication Review-SIGCOMM'10(ACM of Association for Computing Machinery data communication special interest group compunication review-data communication special interest group 10) delivered the article that is entitled as " Helios:a hybrid electrical/optical switch architecture for modular data centers " on > > the 40th volume the 4th phase 339-350 page, traffic demand method of estimation based on flow velocity has been proposed in article, flow condition by each VOQ in the current high-speed mixing switching system of Real-Time Monitoring is used as the factor that traffic demand is estimated, the advantage of the method is to react more exactly current traffic demand, but its shortcoming is also very obvious, flow rate information can not be reacted the traffic demand of sudden flow, therefore, in the very unsettled situation of flow, can cause the growth of the occupancy of buffer memory, thereby bring packet loss and time delay to a certain degree to rise.Meanwhile, in high speed switching system, the accurate detection of flow velocity also can bring very large overhead.
Summary of the invention
The present invention is directed to the above-mentioned shortcoming and defect that prior art exists, traffic demand method of estimation in a kind of grouping and circuit hybrid switching system is proposed, make the present invention have simple, easy to implement, do not increase additional system expense, estimated result is accurate, and is widely used in various traffic environment.In the present invention, buffer memory takies with flow rate information and participates in the process of whole traffic demand estimation simultaneously; The result of scheduling---resource allocation information feeds back the process that traffic demand is estimated that participates in, and the process that process of scheduling and traffic demand are estimated forms closed feedback system.
The present invention is achieved through the following technical solutions, and the present invention realizes by the length information of existing VOQ buffer queue in conventional hybrid switching system and the resource allocation information of this VOQ being carried out to the process that intelligence sample, information extraction and estimation, traffic demand estimate.Wherein, VOQ queue length information refer to input/output port between because certain reason is buffered in the grouping number of input port; VOQ resource allocation information refers to system assignment to the number of resources (as timeslot number, number of wavelengths) of this queue for exchanging, and comprises packet switched portion and circuit switching part; Intelligence sample refers to the process that the information of this queue is periodically obtained and stored; Information extraction and estimation comprise extracts VOQ queue increasing value, estimation VOQ flow rate information two parts, wherein, the extraction of VOQ queue increasing value obtains by VOQ queue length sampled value, and it is by existing resource allocation information and processed queue increasing value, to go to estimate the flow rate information of VOQ that VOQ flow rate information is estimated; Traffic demand estimates it is that the estimation that all information by sampling, extract, estimating are carried out total traffic demand operates above, and processes the steady component that obtains traffic demand and distribute for circuit.
Traffic demand method of estimation according in grouping provided by the invention and circuit hybrid switching system, comprises following process:
Intelligence sample process: the length the record that periodically read VOQ;
Information extraction and estimation procedure: according to the length of VOQ, the increasing value of VOQ extracted and the flow velocity of this VOQ is estimated, obtaining estimated result;
Traffic demand estimation procedure: according to the estimated result obtaining in information extraction and estimation procedure, total traffic demand is estimated, and extracted the process that the steady component in total flow needs estimate value distributes for circuit.
Preferably, specifically comprise the steps:
Step 1, periodic samples VOQ length, wherein, the control plane of switching system is sampled by periodically reading VOQ length information from hardware layer;
The increasing value of step 2, extraction VOQ, wherein, whenever the sampling completing in step 1 VOQ length, by doing calculus of differences with last sampled value, this value, for timing, is positive growth, on the contrary, for negative growth, thereby calculate this VOQ increasing value, key-course records storage to this increasing value;
Step 3, the flow velocity of VOQ is estimated, wherein, exchange resource quantity by the control plane of switching system is distributed to this VOQ is (as timeslot number, number of wavelengths) and the VOQ increasing value obtaining in step 2 be added, sum is divided by with the sampling interval again, obtains the flow velocity estimated value to this VOQ;
Step 4, by flow velocity estimated value and VOQ length estimated flow demand, wherein, whole traffic demand is divided into two parts, part A comes from newly arrived grouping bag, part B comes from the grouping bag of buffer memory, the traffic demand of newly arrived grouping bag is embodied on flow velocity, and the demand of this part A flow velocity estimated value that directly use is obtained by step 3 is as the needs estimate value of part A; The traffic demand of the grouping bag of buffer memory is embodied in queue length, and the demand of this part B is directly proportional to the VOQ length that sampling in step 1 obtains, and is about to be multiplied by the VOQ length of proportionality coefficient c as the needs estimate value of part B; Whole flow needs estimate value is obtained by the weighting summation of the needs estimate value of part A and part B;
The traffic demand that step 5, calculating are used for distributor circuit, wherein, carries out the record of short-term to the whole flow needs estimate value calculating in step 4, and does and extract the operation (as average, filtering) of stablizing component; The whole flow estimated value obtaining is participated in circuit directly and is distributed.
Preferably, according to the resource allocation information of the sampled value of VOQ length and this VOQ, obtain the flow velocity estimated value of this VOQ.
Preferably, proportionality coefficient c value is R/B
q, wherein R is wire rate (line rate), B
qthe maximum queue length that can reach for this VOQ.
Compared with prior art, the present invention has following beneficial effect:
1) by length of buffer queue, information gathering completes whole estimation procedure, the support that does not need hardware layer to provide flow velocity to detect in real time, the realization of can upgrading on existing hybrid switching system, thus reduce costs.
2) resource allocation information, as the result of whole scheduling, feeds back and participates in traffic demand estimation, is the process of a feedback iteration, and result is more accurate.
3) by consider that buffer memory takies and flow velocity estimated value simultaneously, catch the systematic jitters factor that burst flow and regime flow bring, applicable traffic environment is wider, has improved stability and the applicability of system simultaneously.
Accompanying drawing explanation
By reading the detailed description of non-limiting example being done with reference to the following drawings, it is more obvious that other features, objects and advantages of the present invention will become:
Fig. 1 is the flow chart that in the embodiment of the present invention, hybrid switching system carries out traffic demand estimation.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art further to understand the present invention, but not limit in any form the present invention.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, can also make some distortion and improvement.These all belong to protection scope of the present invention.
Key of the present invention is VOQ queue length to gather, and complete the estimation to flow velocity in conjunction with resource allocation information, finally by the traffic demand by the grouping packet generation of buffer memory with by the traffic demand of new arrival grouping packet generation, integrate and obtain total traffic demand, and the steady component extracting wherein distributes for circuit.Because it is tightr that traffic demand is combined with scheduling of resource, thus in the present embodiment by the two in conjunction with expansion.In conjunction with Fig. 1, concrete methods of realizing is as follows:
Step S101: from hardware layer, read VOQ length information, and record.
The control plane of hybrid switching system obtains the queue length of each VOQ of this input port by reading input port cache information, and record is stored in the register of key-course.
Step S102: extract queue increasing value according to queue length information, and in conjunction with the resource allocation information of this queue, estimate the flow velocity of this queue.
Read one constantly VOQ queue length register in value, and do difference operation with the VOQ queue length of current time, obtain VOQ queue increasing value, the exchange resource quantitative value of distributing to this VOQ with current time control plane is added, and be divided by with the sampling interval, obtain the flow velocity estimated value of this VOQ, and be stored in register.
Wherein, unified in queue increasing value and resource apportioning cost Xu unit.That is what, queue increasing value was described is the grouping bag quantity that in the sampling interval, queue length increases; What resource apportioning cost was described is within the sampling interval, to distribute to the resource quantity of this VOQ for exchanging, and its numerical value is converted to this resource and has been used for the grouping bag quantity of exchange.
Step S103: according to flow velocity estimated value and queue length, estimate total traffic demand.
Whole traffic demand is divided into two parts, part A comes from newly arrived grouping bag, and part B comes from the grouping bag of buffer memory, and the traffic demand of newly arrived grouping bag directly can reflect from flow velocity, therefore, this part demand is numerically equal to flow velocity estimated value.The traffic demand of grouping bag and the queue length of VOQ of buffer memory become positive correlation, and proportionality coefficient c value is R/B in this embodiment
q, wherein R is wire rate (line rate), B
qthe maximum queue length that can reach for this VOQ, thus the traffic demand value of buffered packet bag can be obtained.Finally, the needs estimate value that part A, B is corresponding is weighted addition, can obtain total traffic demand estimated value.Wherein, in the present embodiment, two-part weight factor all gets 1, and this value is relevant with system, if system cache is very little, wire rate is very large, the weight factor of buffer memory part should be as far as possible little, and the weight factor of wire rate part should be tried one's best greatly.Otherwise, do inverse operation.
Step S104: extract the steady component of total flow demand, the foundation of distributing as circuit.
In this embodiment, the traffic demand value estimating is carried out to the record of short-term, then get that it is average, obtain the steady component of VOQ traffic demand, and the foundation in order to distribute as circuit.
Step S105: according to the available exchange resource number of system and the traffic demand that estimated, the adjustment before divide into groups/circuit allocation.
Because weighting repeated factor in step S103 is 1, therefore, total traffic demand estimated value can surpass the bandwidth that switching node can provide when duty ratio is larger.In this embodiment, successively the traffic demand of the traffic demand of input port and output port is done to percentage adjustment, the foundation that the traffic demand value of finally usining after adjusting is distributed as circuit, distributes the circuit of hybrid switching system.Remove the exchange resource after circuit distributes, for carrying out packet switching.
The present embodiment makes hybrid switching system more reasonable and efficient to the distribution of exchange resource, do not need extra hardware layer support, so overhead is little, is easy to realize.Meanwhile, owing to considering affecting the parameter of hybrid switching system performance simultaneously, make the method there is applicability widely, can tackle extreme traffic environment.
Above specific embodiments of the invention are described.It will be appreciated that, the present invention is not limited to above-mentioned particular implementation, and those skilled in the art can make various distortion or modification within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (4)
1. the traffic demand method of estimation in grouping and circuit hybrid switching system, is characterized in that, comprises following process:
Intelligence sample process: the length the record that periodically read VOQ;
Information extraction and estimation procedure: according to the length of VOQ, the increasing value of VOQ extracted and the flow velocity of this VOQ is estimated, obtaining estimated result;
Traffic demand estimation procedure: according to the estimated result obtaining in information extraction and estimation procedure, total traffic demand is estimated, and extracted the process that the steady component in total flow needs estimate value distributes for circuit.
2. the traffic demand method of estimation in grouping according to claim 1 and circuit hybrid switching system, is characterized in that, specifically comprises the steps:
Step 1, periodic samples VOQ length, wherein, the control plane of switching system is sampled by periodically reading VOQ length information from hardware layer;
The increasing value of step 2, extraction VOQ, wherein, whenever the sampling completing in step 1 VOQ length, by doing calculus of differences with last sampled value, thereby calculate this VOQ increasing value, key-course records storage to this increasing value;
Step 3, the flow velocity of VOQ is estimated, wherein, by the control plane of switching system being distributed to the VOQ increasing value obtaining in the exchange resource quantity of this VOQ and step 2, be added, sum is divided by with the sampling interval again, obtains the flow velocity estimated value to this VOQ;
Step 4, by flow velocity estimated value and VOQ length estimated flow demand, wherein, whole traffic demand is divided into two parts, part A comes from newly arrived grouping bag, part B comes from the grouping bag of buffer memory, the traffic demand of newly arrived grouping bag is embodied on flow velocity, and the demand of this part A flow velocity estimated value that directly use is obtained by step 3 is as the needs estimate value of part A; The traffic demand of the grouping bag of buffer memory is embodied in queue length, and the demand of this part B is directly proportional to the VOQ length that sampling in step 1 obtains, and is about to be multiplied by the VOQ length of proportionality coefficient c as the needs estimate value of part B; Whole flow needs estimate value is obtained by the weighting summation of the needs estimate value of part A and part B;
The traffic demand that step 5, calculating are used for distributor circuit, wherein, carries out the record of short-term to the whole flow needs estimate value calculating in step 4, and does and extract the operation of stablizing component; The whole flow estimated value obtaining is participated in circuit directly and is distributed.
3. the traffic demand method of estimation in grouping according to claim 2 and circuit hybrid switching system, it is characterized in that, according to the resource allocation information of the sampled value of VOQ length and this VOQ, obtain the flow velocity estimated value of this VOQ.
4. the traffic demand method of estimation in grouping according to claim 2 and circuit hybrid switching system, is characterized in that, proportionality coefficient c value is R/B
q, wherein R is wire rate (line rate), B
qthe maximum queue length that can reach for this VOQ.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310445946.4A CN103546337A (en) | 2013-09-25 | 2013-09-25 | Flow demand estimation method for packet and circuit hybrid switching system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310445946.4A CN103546337A (en) | 2013-09-25 | 2013-09-25 | Flow demand estimation method for packet and circuit hybrid switching system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103546337A true CN103546337A (en) | 2014-01-29 |
Family
ID=49969412
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310445946.4A Pending CN103546337A (en) | 2013-09-25 | 2013-09-25 | Flow demand estimation method for packet and circuit hybrid switching system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103546337A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105553881A (en) * | 2015-12-15 | 2016-05-04 | 上海交通大学 | General method for network resource allocation in hybrid switching system |
WO2018233560A1 (en) * | 2017-06-20 | 2018-12-27 | 华为技术有限公司 | Dynamic scheduling method, device, and system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020154649A1 (en) * | 2001-02-23 | 2002-10-24 | Masayuki Takase | Packet switching system |
CN1487699A (en) * | 2002-09-30 | 2004-04-07 | 华为技术有限公司 | Flow rate simulating method and device for exchange network test |
CN102611605A (en) * | 2011-01-20 | 2012-07-25 | 华为技术有限公司 | Scheduling method, device and system of data exchange network |
-
2013
- 2013-09-25 CN CN201310445946.4A patent/CN103546337A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020154649A1 (en) * | 2001-02-23 | 2002-10-24 | Masayuki Takase | Packet switching system |
CN1487699A (en) * | 2002-09-30 | 2004-04-07 | 华为技术有限公司 | Flow rate simulating method and device for exchange network test |
CN102611605A (en) * | 2011-01-20 | 2012-07-25 | 华为技术有限公司 | Scheduling method, device and system of data exchange network |
Non-Patent Citations (1)
Title |
---|
PINGQING LI, ET AL: "Traffic demand estimation for hybrid switching systems", 《TRANSPARENT OPTICAL NETWORKS (ICTON),2013 15TH INTERNATIONAL CONFERENCE ON》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105553881A (en) * | 2015-12-15 | 2016-05-04 | 上海交通大学 | General method for network resource allocation in hybrid switching system |
CN105553881B (en) * | 2015-12-15 | 2018-10-23 | 上海交通大学 | The universal method that Internet resources distribute in hybrid switching system |
WO2018233560A1 (en) * | 2017-06-20 | 2018-12-27 | 华为技术有限公司 | Dynamic scheduling method, device, and system |
US11121979B2 (en) | 2017-06-20 | 2021-09-14 | Huawei Technologies Co., Ltd. | Dynamic scheduling method, apparatus, and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102611605B (en) | Scheduling method, device and system of data exchange network | |
TW200830786A (en) | Methods, systems, and computer program products for grooming traffic on a shared network facility based on a peak utilization analysis of the shared network facility | |
CN108491255B (en) | Self-service MapReduce data optimal distribution method and system | |
US11171869B2 (en) | Microburst detection and management | |
CN103188152B (en) | Cognitive network quality of service (QoS) routing method based on service differentiation | |
WO2011089899A1 (en) | Semiconductor system, relay apparatus, and chip circuit | |
CN103004156B (en) | Service aware rate throttling of delay tolerant traffic for energy efficient routing | |
CN110474842A (en) | A kind of communication means, customer terminal equipment and transmission network | |
CN107454009B (en) | Data center-oriented offline scene low-bandwidth overhead traffic scheduling scheme | |
US10153952B2 (en) | Network traffic monitoring system and method thereof | |
CN113132265B (en) | Multi-stage scheduling method and device for multi-path Ethernet | |
US9344384B2 (en) | Inter-packet interval prediction operating algorithm | |
CN102946443B (en) | Multitask scheduling method for realizing large-scale data transmission | |
CN103546337A (en) | Flow demand estimation method for packet and circuit hybrid switching system | |
CN108390737B (en) | Single satellite link performance analysis method based on intermittent service capability model | |
CN114257554B (en) | Scheduling method for improving bandwidth utilization rate of BE (flow element) stream of TSN (traffic channel network) | |
JP2003234767A (en) | Real time estimation of equivalent bandwidth utilization | |
CN105049365A (en) | Adaptive frequency modulation energy-saving method for multi-core multi-thread intrusion detection device | |
CN106209683B (en) | Data transmission method and system based on data center's wide area network | |
CN115277504A (en) | Network traffic monitoring method, device and system | |
Chi et al. | Queuing theory based service performance evaluation under H2h and M2 M blending traffic arriving | |
Mo et al. | C-INT: An efficient cluster based in-band network telemetry | |
US20220385559A1 (en) | Topology estimation system, traffic generation apparatus, and traffic generation method | |
Moltafet et al. | Average age of information for a multi-source M/M/1 queueing model with packet management and self-preemption in service | |
KR100333475B1 (en) | Rate proportional self-clocked fair queueing apparatus and method for high-speed packet-switched networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140129 |