CN109088795B - Controller performance analysis method based on equal-interval sampling - Google Patents

Controller performance analysis method based on equal-interval sampling Download PDF

Info

Publication number
CN109088795B
CN109088795B CN201811079839.3A CN201811079839A CN109088795B CN 109088795 B CN109088795 B CN 109088795B CN 201811079839 A CN201811079839 A CN 201811079839A CN 109088795 B CN109088795 B CN 109088795B
Authority
CN
China
Prior art keywords
switch
network
controller
time
data flow
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.)
Active
Application number
CN201811079839.3A
Other languages
Chinese (zh)
Other versions
CN109088795A (en
Inventor
蒋定德
林浩杰
张溢航
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201811079839.3A priority Critical patent/CN109088795B/en
Publication of CN109088795A publication Critical patent/CN109088795A/en
Application granted granted Critical
Publication of CN109088795B publication Critical patent/CN109088795B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/02—Capturing of monitoring data
    • H04L43/022—Capturing of monitoring data by sampling
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/02—Capturing of monitoring data
    • H04L43/026—Capturing of monitoring data using flow identification
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters

Abstract

The invention discloses a controller performance analysis method based on equal-interval sampling, and belongs to the technical field of future networks. The invention can systematically and comprehensively analyze and compare the performances of different controllers. The method comprises the following specific steps: the method comprises the following steps: initializing a network, inputting relevant parameters, and selecting a source node and a destination node; step two: selecting a source switch and a destination switch by using a switch selection algorithm so as to measure data flow information of the source switch and the destination switch to calculate a performance index of the network; step three: in a period of time, the controller sends a data flow statistic request to the source switch and the target switch at regular time; step four: measuring information of the data stream corresponding to the data stream statistics request in the step three; step five: and calculating network performance indexes, such as network packet loss rate, network delay, throughput, delay jitter and measurement error, and comprehensively measuring the performance characteristics of the controller when a data packet appears for the first time by calculating the network performance indexes.

Description

Controller performance analysis method based on equal-interval sampling
Technical Field
The invention belongs to the technical field of future networks, and particularly relates to a controller performance analysis method based on equal-interval sampling.
Background
Currently, with the continuous expansion of network scale, the traditional network architecture is difficult to meet the needs of current operators, enterprises and users. In order to effectively manage a complex network, a Software Defined Network (SDN) architecture having a feature of separating a data plane from a control plane is gradually accepted, and this feature can effectively simplify network management. In the SDN paradigm, the data plane is greatly simplified by the feature of separation of the control plane and the data plane, and both the complex processing and calculation processes are handed over to the control plane. As a core component of the control plane, the controller can be regarded as an important foundation because it can perform centralized control management on network traffic according to the policies of network operators and software developers. It can be seen that the performance of the controller determines the extensibility of the software defined network, which plays an important role in the architecture of the software defined network.
Currently, the implementation of controllers has become diversified, and the industry and the academia have successively developed different kinds of controllers, which have implemented different programming languages and function sets, and have considerable differences from the original design prototype to the controllers. Therefore, how to accurately and efficiently evaluate the performance of the controller in the software defined network is a significant challenge facing the current SDN field.
SDN is a new network architecture. The architecture is characterized by a separation of the data plane from the control plane and by the ability to be directly controlled by the program. In existing networks, control and forwarding of traffic is performed entirely by network devices (e.g., switches, routers), while SDN separates control functions from network devices and performs centralized network control. The data plane contains network forwarding elements (switches and routers) and the control plane contains controllers. The controller utilizes a control-forwarding communication interface for centralized control of network forwarding units on the data plane and provides flexible programmability, which is not present in existing networks. Therefore, SDN completely changes the existing network architecture, and the controller is an essential component of the SDN architecture, which will contribute to the development of SDN and make it necessary to evaluate and compare the existing SDN controller. However, there is currently no systematic and comprehensive analysis method for such a large number of SDN controllers. Due to lack of sufficient information, researchers only select controllers to complete their own research by experience or indirect data, however, this lack of a strong method makes SDN controller resources not fully utilized.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a more comprehensive and systematic measuring method for systematically analyzing and comparing the current mainstream controller.
The technical scheme of the invention is a controller performance analysis method based on equal-interval sampling, which comprises the following steps:
step 1: network initialization
Initializing a network, and randomly selecting a source node and a destination node from a network topology;
step 2: selective switch
Selecting a source switch and a target switch by using a switch selection algorithm, measuring data flow information of the selected source switch and the selected target switch and calculating a performance index of the controller;
and step 3: sending data flow statistics request
In a period of time, the controller sends flow statistic requests to the source switch and the target switch at regular time;
and 4, step 4: measuring data flow information
Measuring flow information corresponding to the data flow statistical request sent in the step 3;
and 5: computing network performance indicators
And calculating the network packet loss rate, the network delay, the throughput, the delay jitter and the measurement error, and comprehensively measuring the performance characteristics of the controller when the data packet occurs for the first time by calculating the network performance indexes.
Further, the network initialization in step 1 includes, specifically: the method comprises the steps of a controller, a non-edge switch connected with the controller, an edge switch connected with the controller, a host connected with the edge switch, an active flow currently sent by the host, a source node, a destination node, a measurement period of the switch, and an upper measurement time limit between the source switch and the destination switch.
Further, the switch selection process in step 2 specifically includes the following steps:
traversing the whole monitoring path, and if the edge switch is a source node and is not a destination node, selecting the switch as a source switch; if the edge switch is a destination node and not a source node, then this switch is selected as the destination switch.
Further, the sending of the data flow statistics request in step 3 specifically includes the following steps: and in a specified time, starting from the current measurement time, the controller respectively sends flow counting requests to the source switch and the target switch at intervals.
Further, the data flow information is measured in the step 4, and the specific process is as follows:
and (3) capturing the data flow which flows through the switch in the network and is matched with the data flow corresponding to the flow counting request sent by the controller in the step (3) by the switch according to the flow counting request sent by the controller in the step (3) through a flow matching rule and a matching algorithm, obtaining the statistical information of the data flow and transmitting the measurement information back to the controller.
Further, the network performance index calculated in step 5 specifically includes the following processes:
and 4, calculating packet loss rate, time delay, link utilization rate and time delay jitter according to the data flow information measured in the step 4, and comprehensively measuring the performance characteristics of the controller when the data packet appears for the first time by calculating the network performance indexes.
The specific calculation formula is as follows:
(1) packet loss rate li：
Cfirst＝Cfirst_j-Cfirst_(j-i) (1)
Cfinal＝Cfinal_j-Cfinal_(j-i) (2)
Li＝Cfirst-Cfinal (3)
Figure BDA0001801649190000031
Wherein, Cfirst_jRepresenting the number of packets of the data flow of the counter in the first switch recorded at the moment j; cfirst_(j-i)Representing the number of packets of the data flow of the counter in the first switch recorded at the time j-i; cfinal_jThe number of packets of the data flow of the counter in the last switch is recorded at the moment j; cfinal_(j-i)Representing the number of packets of the data flow of the counter in the last switch recorded at the time j-i; l isiA difference value representing the number of packets of the incremental data flow of the counters in the first switch and the last switch during the measurement period i;
(2) network delay tdelay：
T1＝Tstart_1-Tleave_2 (5)
T2＝Tstart_2-Tleave_1 (6)
Figure BDA0001801649190000032
Wherein, T1Representing the time difference of the two switches for processing the data stream; t isstart_1Represents the start time of the data flow through switch 1 in the network; t isleave_2Represents the time at which the data stream leaves switch 2 in the network; t is2Representing the time difference of the two switch processing machines; t isstart_2Represents the start time of the data flow through switch 2 in the network; t isleave_1Represents the time when the data stream leaves switch 1 in the network; t isc-s1And Tc-s2Representing controller round trip times to switch 1 and switch 2, respectively;
(3) throughput Ui：
Figure BDA0001801649190000033
Figure BDA0001801649190000034
Figure BDA0001801649190000035
Figure BDA0001801649190000036
Figure BDA0001801649190000041
Wherein, Cfirst_avgA packet number representing an average data flow of a counter in the first switch during a measurement period i; cfinal_avgTo representIn the measurement period i, the packet number of the average data flow of the counter in the last switch; u shapefirstRepresenting the throughput through the first switch during the measurement period i; u shapefinalRepresents the throughput through the last switch during the measurement period i; u shapeiRepresents the average throughput through the first switch and the last switch during the measurement period i;
(4) delay Jitter:
Figure BDA0001801649190000042
where n represents the number of delay sequences of the measurement network, td-jRepresenting the j-th sequence of delays in measuring the network, Avg (t)d-j) Representing the average value of the time delay sequence of the measurement network;
(5) measurement error rmse (x):
Figure BDA0001801649190000043
wherein x isiThe value of the data stream rate for the ith equally spaced sample,
Figure BDA0001801649190000044
the true value of the dataflow rate for the ith host, n is the number of samples.
The invention has the beneficial effects that:
the invention provides a controller performance analysis method based on equal-interval sampling. The method utilizes a switch selection algorithm and an equidistant sampling method to systematically analyze and compare the SDN controllers (a POX controller, a FloodLight controller, an RYU controller, an ONOS controller and an OpenDayLight controller) which are mainstream at present, provides a more comprehensive and systematic measuring method, comprehensively measures the performance characteristics of the controllers when a data packet appears for the first time by calculating the network performance indexes, and quantitatively analyzes the performance indexes of the controllers so as to provide a practical and effective research scheme for future researchers. It can be seen that the present invention compares controller performance using only equally spaced sampling, which is an advantage of the analysis method of the present invention.
More importantly, the invention carries out quantitative analysis on the performance index of the controller. In the measurement, a stable analysis method of sampling at equal intervals is adopted, so that the measurement errors of the obtained data streams are basically consistent, the performance analysis of the controller is not influenced by the measurement method, and the measurement accuracy of the data streams in the network is also ensured. The experimental result shows that the performance index of the controller measured by the equal-interval sampling method basically has no influence on the performance of the analysis controller, and meanwhile, the measurement accuracy is also ensured.
Drawings
Figure 1 is an SDN measurement mechanism used by the present invention;
figure 2 is an SDN multi-flow measurement mechanism used by the present invention;
FIG. 3 is a diagram of network delay components involved in the present invention;
FIG. 4 is a linear network topology of an example application of the present invention;
FIG. 5 is a mesh-type network topology of an example application of the present invention;
FIG. 6 shows simulation results of an embodiment of the present invention;
(a) average packet loss rate for network topology 1
(b) Average packet loss rate for network topology 2
(c) Average delay for network topology 1
(d) Average delay for network topology 2
(e) Average throughput for network topology 1
(f) Average throughput for network topology 2
(g) Average delay jitter for network topology 1
(h) Average delay jitter for network topology 2
(i) Is the root mean square error.
Detailed Description
And establishing a network topology structure by using a Mininet simulator, and connecting the network topology structure with the controller. Two network simulation topologies are implemented herein: linear network topologies and mesh network topologies, as shown in fig. 4 and 5, respectively; and analyzing and comparing the performance indexes of the controller by using the method provided by the document through the established network topology structure. The controller performance analysis method based on equal-interval sampling comprises the following specific steps:
the method comprises the following steps: network initialization
Network initialization, namely randomly selecting a source node and a destination node from a network topology; in an embodiment of the invention, two different topologies are applied. Fig. 4 is a simple linear network topology structure, which is composed of 1 controller, 2 switches, and 4 hosts, and finally forms a bus-type network with 2 network nodes and 4 host nodes; fig. 5 is a mesh-type network topology, which is composed of 1 controller, 6 switches and 6 hosts, wherein the switches are connected to each other, and finally a mesh-type network of 6 network nodes and 6 host nodes is formed.
Step two: selective switch
Selecting corresponding source node and destination node switches, namely a first switch and a last switch mentioned in the method of the present invention, and selecting S1 and S2 as the first switch and the last switch in the linear topology as shown in fig. 4; meanwhile, in the mesh type network topology as shown in fig. 5, S1 and S6 are selected as the first switch and the last switch, and relevant data flow information is measured;
step three: sending data flow statistics request
The controller periodically sends flow statistics requests to the source switch and the destination switch over a period of time. For the linear topology, as shown in fig. 4, only data flows with IP addresses of host h1 in the switches S1 and S2 are counted; for the mesh topology as shown in fig. 5, only the data flow with IP address of host h1 in the switches S1 and S6 is counted, which is also the requirement of the method of the present invention.
Measuring data flow information
In the last step, the controller periodically sends flow statistics requests to the source switch and the destination switch, i.e., the S1 and S2 switches in fig. 4, the S1 and S6 switches in fig. 5; after the switch receives the flow statistic request of the controller, the corresponding flow is collected and subjected to statistic analysis processing through the flow matching rule and the matching algorithm, the statistic result of the corresponding flow is sent to the controller, and the controller collects flow statistic information from each measuring switch.
Step five: computing network performance indicators
And calculating the network packet loss rate, the network delay, the throughput, the delay jitter and the measurement error, and comprehensively measuring the performance characteristics of the controller when the data packet occurs for the first time by calculating the network performance indexes. The results are shown in FIG. 6.

Claims (1)

1. A controller performance analysis method based on equal-interval sampling comprises the following specific steps:
step 1: network initialization
Initializing a network, and randomly selecting a source node and a destination node from a network topology;
the network initialization includes the following specific contents: the method comprises the following steps that a controller, a non-edge switch connected with the controller, an edge switch connected with the controller, a host connected with the edge switch, an active flow currently sent by the host, a source node, a destination node, a measurement period of the switch and a measurement time upper limit between the source switch and the destination switch are controlled;
step 2: selective switch
Selecting a source switch and a target switch by utilizing a switch selection algorithm so as to measure the data flow information of the selected source switch and the target switch and calculate the performance index of the network; traversing the whole monitoring path, and if the edge switch is a source node and is not a destination node, selecting the switch as a source switch; if the edge switch is a destination node and not a source node, selecting the switch as a destination switch;
and step 3: sending data flow statistics request
In a period of time, the controller sends flow statistic requests to the source switch and the target switch at regular time; in a specified time, starting from the current measurement time, the controller respectively sends flow counting requests to the source switch and the target switch at intervals;
and 4, step 4: measuring data flow information
Measuring information of the data stream corresponding to the data stream statistical request in the step 3; through a flow matching rule and a matching algorithm, according to the flow counting request sent by the controller in the step 3, the switch captures a data flow which flows through the switch in the network and is matched with the data flow corresponding to the flow counting request sent by the controller in the step 3, obtains the statistical information of the data flow and transmits the measurement information back to the controller;
and 5: computing network performance indicators
Calculating packet loss rate, time delay, link utilization rate and time delay jitter according to the data flow information measured in the step 4, and comprehensively measuring the performance characteristics of the controller when the data packet appears for the first time by calculating the network performance indexes;
the specific calculation formula is as follows:
(1) packet loss rate li：
Cfirst＝Cfirst_j-Cfirst_(j-i) (1)
Cfinal＝Cfinal_j-Cfinal_(j-i) (2)
Li＝Cfirst-Cfinal (3)
Figure FDA0003258528860000011
Wherein, Cfirst_jRepresenting the number of packets of the data flow of the counter in the first switch recorded at the moment j; cfirst_(j-i)Representing the number of packets of the data flow of the counter in the first switch recorded at the time j-i; cfinal_jThe number of packets of the data flow of the counter in the last switch is recorded at the moment j; cfinal_(j-i)Representing the number of packets of the data flow of the counter in the last switch recorded at the time j-i; l isiIs shown inMeasuring the packet number difference of the incremental data flows of the counters in the first switch and the last switch in the time period i;
(2) network delay tdelay：
T1＝Tstart_1-Tleave_2 (5)
T2＝Tstart_2-Tleave_1 (6)
Figure FDA0003258528860000021
Wherein, T1Representing the time difference of the two switches for processing the data stream; t isstart_1Represents the start time of the data flow through switch 1 in the network; t isleave_2Represents the time at which the data stream leaves switch 2 in the network; t is2Representing the time difference of the two switch processing machines; t isstart_2Represents the start time of the data flow through switch 2 in the network; t isleave_1Represents the time when the data stream leaves switch 1 in the network; t isc-s1And Tc-s2Representing controller round trip times to switch 1 and switch 2, respectively;
(3) throughput Ui：
Figure FDA0003258528860000022
Figure FDA0003258528860000023
Figure FDA0003258528860000024
Figure FDA0003258528860000025
Figure FDA0003258528860000026
Wherein, Cfirst_avgA packet number representing an average data flow of a counter in the first switch during a measurement period i; cfinal_avgA packet number representing an average data flow of a counter in the last switch during the measurement period i; u shapefirstRepresenting the throughput through the first switch during the measurement period i; u shapefinalRepresents the throughput through the last switch during the measurement period i; u shapeiRepresents the average throughput through the first switch and the last switch during the measurement period i;
(4) delay Jitter:
Figure FDA0003258528860000031
where n represents the number of delay sequences of the measurement network, td-jRepresenting the j-th sequence of delays in measuring the network, Avg (t)d-j) Representing the average value of the time delay sequence of the measurement network;
(5) measurement error rmse (x):
Figure FDA0003258528860000032
wherein x isiThe value of the data stream rate for the ith equally spaced sample,
Figure FDA0003258528860000033
the true value of the dataflow rate for the ith host, n is the number of samples.
CN201811079839.3A 2018-09-17 2018-09-17 Controller performance analysis method based on equal-interval sampling Active CN109088795B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811079839.3A CN109088795B (en) 2018-09-17 2018-09-17 Controller performance analysis method based on equal-interval sampling

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811079839.3A CN109088795B (en) 2018-09-17 2018-09-17 Controller performance analysis method based on equal-interval sampling

Publications (2)

Publication Number Publication Date
CN109088795A CN109088795A (en) 2018-12-25
CN109088795B true CN109088795B (en) 2021-11-02

Family

ID=64841671

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811079839.3A Active CN109088795B (en) 2018-09-17 2018-09-17 Controller performance analysis method based on equal-interval sampling

Country Status (1)

Country Link
CN (1) CN109088795B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103795596A (en) * 2014-03-03 2014-05-14 北京邮电大学 Programmable control SDN measuring system and method
CN105357068A (en) * 2015-11-03 2016-02-24 华中科技大学 OpenFlow network flow control method for QoS assurance of application
CN105515793A (en) * 2014-09-23 2016-04-20 中国电信股份有限公司 On-line charging processing method and system, Openflow controller and Openflow switch

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103795596A (en) * 2014-03-03 2014-05-14 北京邮电大学 Programmable control SDN measuring system and method
CN105515793A (en) * 2014-09-23 2016-04-20 中国电信股份有限公司 On-line charging processing method and system, Openflow controller and Openflow switch
CN105357068A (en) * 2015-11-03 2016-02-24 华中科技大学 OpenFlow network flow control method for QoS assurance of application

Also Published As

Publication number Publication date
CN109088795A (en) 2018-12-25

Similar Documents

Publication Publication Date Title
Huo et al. A SDN‐based fine‐grained measurement and modeling approach to vehicular communication network traffic
Jain et al. Applying big data technologies to manage QoS in an SDN
KR101234326B1 (en) Distributed traffic analysis
CN107896160B (en) A kind of data center network flowmeter factor method based on distributed system
WO2013038279A1 (en) Network-wide flow monitoring in split architecture networks
WO2011130957A1 (en) Method and apparatus for online distinguishing transmission control protocol traffic by using data flow head characteristics
Liu et al. A queueing-theoretic foundation of available bandwidth estimation: single-hop analysis
Xu et al. Lightweight and adaptive service api performance monitoring in highly dynamic cloud environment
CN109587000B (en) High-delay anomaly detection method and system based on crowd-sourcing network measurement data
Javed et al. A stochastic model for transit latency in OpenFlow SDNs
CN116545936A (en) Congestion control method, system, device, communication equipment and storage medium
Liu et al. A SDN-based active measurement method to traffic QoS sensing for smart network access
CN111147323A (en) Speed measuring method and device
CN109088795B (en) Controller performance analysis method based on equal-interval sampling
Middleton et al. Scalable classification of QoS for real-time interactive applications from IP traffic measurements
Harkous et al. Modeling and performance analysis of P4 programmable devices
Ogasawara et al. Performance analysis of traffic classification in an OpenFlow switch
CN104363142A (en) Automatic data center network performance bottleneck analysis method
CN115695216A (en) Big data analysis method for internet traffic flow direction
CN115022191A (en) Quick inversion method for end-to-end flow in IPv6 network
Bianco et al. OpenFlow driven ethernet traffic analysis
Bolodurina et al. Comprehensive approach for optimization traffic routing and using network resources in a virtual data center
CN103259731A (en) Method for generating and simplifying self-similarity traffic of network key node based on opening/closing source model
Pekar et al. Towards threshold‐agnostic heavy‐hitter classification
Tang et al. Intelligent Awareness of Delay-Sensitive Internet Traffic in Digital Twin Network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant