Lightweight collaborative network flow measurement method and system oriented to data center network

Abstract

The invention relates to the technical field of network measurement, and particularly provides a lightweight collaborative network flow measurement method and system for a data center network, and the method comprises the steps: determining a to-be-measured network flow as a first network flow or a second network flow, and enabling the first network flow to be larger than the second network flow; under the condition that the to-be-measured network flow is the first network flow, the measurement task is uniformly distributed to each switch in a first path, and the first path is a transmission path of the to-be-measured network flow; when the to-be-measured network flow is a second network flow, the measurement task is allocated to a preset switch, the preset switch is any switch in a second path, and the second path is a transmission path of the to-be-measured network flow; controlling the switch to measure and record the flow information of the network flow to be measured according to the distributed measurement task; and responding to the query instruction, acquiring flow information, and calculating the number of data packets in the network flow to be detected. According to the invention, the data packet processing efficiency and the overall measurement precision are high.

Description

Technical Field

[0001] The invention relates to the field of network measurement technology, and in particular to a lightweight collaborative network flow measurement method and system for a data center network. Background Art

[0002] Fine-grained network traffic monitoring can monitor and manage network traffic in server clusters in real time, optimize network performance, and improve user experience. It can also effectively prevent common network attacks and ensure user data security. In addition, efficient fine-grained network traffic monitoring can also optimize resource allocation in the entire network and reduce network operating costs. Therefore, fine-grained network traffic monitoring has become an indispensable management method for network companies.

[0003] As the core hub of network services, data centers bear the important responsibility of connecting users and services, and carry "massive" network traffic all the time. With the continuous increase in the number of Internet users and the popularity of various online services, the user traffic faced by data centers has increased dramatically. Large network companies such as Google, Facebook, and Amazon have hundreds of millions of users accessing network services through their data centers every day, which poses a great challenge to the stable operation and efficient management of data centers. Therefore, it is not a simple matter to monitor and manage the network traffic of the entire network, especially for the data centers of large network companies. In addition, due to the need to provide real-time network services to a large number of users at the same time, the traffic in data centers often has the important feature of "high speed". The data transmission speed in the data center is very fast. For example, hundreds of Gb need to be transmitted within 1 second. Faced with this high-speed characteristic of network traffic, traditional measurement methods are difficult to capture and record the traffic passing through each switch in a timely manner.

[0004] There are a large number of Sketch-based measurement methods that can efficiently record traffic information, but these methods are limited to single-point measurement and are difficult to expand to full-network traffic measurement. Especially for large data center networks, the huge amount of routing information, link bandwidth limitations, and switch resource limitations will greatly affect the strategy formulation and actual deployment of full-network traffic measurement. Since the centralized controller under the Software Defined Network (SDN) architecture has a global perspective of the entire network, the SDN architecture can be used to uniformly manage and control network traffic. However, in the actual measurement process, since the SDN architecture needs to monitor the status of switches in the network in real time, and frequent collaborative communication between the controller and the switch is required, the architecture itself will bring huge bandwidth consumption to the entire link, which is not an efficient solution.

[0005] Regarding the network traffic measurement solutions in related technologies, it is difficult to balance the packet processing efficiency and the overall measurement accuracy when measuring fine-grained network traffic in a large data center, and no effective solution has been proposed so far. Summary of the invention

[0006] The embodiments of the present invention provide a lightweight collaborative network traffic measurement method and system for data center networks, which at least solves the problem that network traffic measurement solutions in related technologies are difficult to balance data packet processing efficiency and overall measurement accuracy when measuring fine-grained network traffic in large data centers.

[0007] The invention provides a lightweight collaborative network traffic measurement method for a data center network, including:

[0008] Determine that the network flow to be tested is a first network flow or a second network flow, wherein the first network flow is greater than the second network flow;

[0009] In a case where the network flow to be measured is the first network flow, evenly distributing the measurement task of the network flow to be measured to each switch in a first path, wherein the first path is a transmission path of the network flow to be measured;

[0010] In the case where the network flow to be measured is the second network flow, allocating the measurement task of the network flow to be measured to a preset switch, wherein the preset switch is any switch in the second path, and the second path is the transmission path of the network flow to be measured;

[0011] Controlling the switch to measure and record the flow information of the network flow to be measured according to the assigned measurement task;

[0012] In response to the query instruction, the flow information is obtained, and the number of data packets in the network flow to be tested is calculated.

[0013] The lightweight collaborative network traffic measurement method for a data center network provided by an embodiment of the present invention determines that the network flow to be measured is a first network flow or a second network flow, including:

[0014] The detector based on the ingress switch calculates the hash value of the data packet in the network flow to be tested, and stores the hash value in the corresponding bucket;

[0015] When the same data packet appears in the network flow to be tested, control the counter in the bucket corresponding to the first hash value to increase by 1, wherein the first hash value is the same hash value of the same data packet, wherein the identification fields of the same data packets are the same;

[0016] When the hash values ​​of different data packets in the network flow to be tested are the same, a hash conflict occurs, and a counter in a bucket corresponding to a second hash value is controlled to be reduced by 1, wherein the second hash value is the same hash value of the different data packets, wherein the identification fields of the different data packets are different;

[0017] Wherein, when the value of any of the counters is 0, the network flow to be tested is determined to be the first network flow; when the value of none of the counters is 0, the network flow to be tested is determined to be the second network flow.

[0018] The lightweight collaborative network traffic measurement method for a data center network provided by an embodiment of the present invention, after determining that the network flow to be measured is the first network flow, the method further includes:

[0019] Storing the hash value of the network flow to be tested into a target bucket, wherein the target bucket is a bucket whose counter value is 0;

[0020] Reset the value of the counter of the target bucket to 1.

[0021] The lightweight collaborative network traffic measurement method for a data center network provided by an embodiment of the present invention determines that the network flow to be measured is a first network flow or a second network flow, and further includes:

[0022] Before calculating the hash value of the data packet in the network flow to be tested, the detector based on the ingress switch samples the data packet in the network flow to be tested according to a preset probability;

[0023] The detector based on the ingress switch calculates the hash value of the sampled data packet, and stores the hash value of the sampled data packet in a corresponding bucket.

[0024] The invention provides a lightweight collaborative network traffic measurement method for a data center network, and the method further includes:

[0025] In a case where the network flow to be tested is the second network flow, determining a hash interval matching requirement based on a preset network topology structure;

[0026] Wherein, the first Paths of different lengths are denoted as , the path is one or more, the path The length is set to , h is the total number of layers of the switch, the path The hash interval that needs to be maintained is denoted as , the hash interval The matching requirements are:

[0027] ;

[0028] in, , j represents the j-th layer switch in the h-layer switches; , k represents the k-th layer switch in the h-layer switches;

[0029] According to the hash interval matching requirement and the network topology structure, the measurement task of the network flow to be measured is allocated to a preset switch.

[0030] The lightweight collaborative network traffic measurement method for data center networks provided by the present invention is based on a preset network topology structure to determine the hash interval matching requirements, including:

[0031] The hash interval matching requirement is determined based on the network flow quantity threshold C and the preset network topology structure. The network flow quantity threshold C needs to meet the following requirements:

[0032] ;

[0033] in, For the The path through the layer switch The number of For the hash interval Length;

[0034] According to the network flow quantity threshold C, the hash interval matching requirement and the network topology structure, the measurement task of the network flow to be measured is allocated to the preset switch.

[0035] The lightweight collaborative network traffic measurement method for a data center network provided by an embodiment of the present invention controls the switch to record the traffic information of the allocated data packets, including:

[0036] The flow information is recorded separately based on a two-layer collaboration sketch, wherein the two-layer collaboration sketch is configured on each of the switches, and the two-layer collaboration sketch includes two layers of buckets, a first layer bucket B L include Buckets, second layer bucket B S include Buckets, Less than ;

[0037] When the network flow to be tested is the first network flow, based on the first layer bucket B L Record the flow information of the data packets in the network flow to be tested; if the network flow to be tested is the second network flow, based on the second layer bucket B SThe flow information of the data packets in the network flow to be tested is recorded.

[0038] The embodiment of the present invention provides a lightweight collaborative network traffic measurement method for a data center network, wherein the traffic information is a flow label or an element label.

[0039] The lightweight collaborative network traffic measurement system for data center networks provided by the present invention includes:

[0040] The server sends or receives the network flow to be tested;

[0041] The switch forwards the network flow to be tested and processes the network flow to be tested;

[0042] The controller determines that the network flow to be tested is a first network flow or a second network flow, wherein the first network flow is greater than the second network flow;

[0043] The controller, further, when the network flow to be measured is the first network flow, evenly distributes the measurement task of the network flow to be measured to each switch in a first path, wherein the first path is a transmission path of the network flow to be measured; when the network flow to be measured is the second network flow, distributes the measurement task of the network flow to be measured to a preset switch, wherein the preset switch is any switch in a second path, and the second path is a transmission path of the network flow to be measured;

[0044] The controller further controls the switch to measure and record the flow information of the network flow to be measured according to the measurement task assigned to it;

[0045] The controller also responds to the query instruction, obtains the flow information, and calculates the number of data packets in the network flow to be tested.

[0046] The electronic device provided by the embodiment of the invention includes: a processor and a memory storing a program, wherein the program includes instructions, and when the instructions are executed by the processor, the processor executes the lightweight collaborative network traffic measurement method for data center networks.

[0047] The invention provides a lightweight collaborative network flow measurement method and system for data center networks, which solves the problem that it is difficult to balance the data packet processing efficiency and the overall measurement accuracy in the existing network flow measurement scheme when measuring fine-grained network flow in a large data center. By determining whether the network flow to be measured is the first network flow or the second network flow, when the network flow to be measured is the first network flow, the measurement task of the network flow to be measured is evenly distributed to each switch on its transmission path to achieve packet-level measurement load balancing; when the network flow to be measured is the second network flow, the measurement task of the network flow to be measured is distributed to any switch on its transmission path to achieve flow-level measurement load balancing; thereby avoiding the data packets of each network flow to be measured being repeatedly measured by different switches, and keeping the number of data packets that each switch needs to process similar, and the number of network flows to be measured that each switch needs to process similar, so as to ensure that the actual measurement load of each switch is slightly different within the predetermined time, the data packet processing efficiency is high, and the overall measurement accuracy is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0049] Figure 1 It is a step flow chart of a lightweight collaborative network traffic measurement method for a data center network in an embodiment of the present invention.

[0050] Figure 2 It is a schematic diagram of the structure of TC Sketch in the embodiment of the present invention.

[0051] Figure 3 It is a schematic diagram of a lightweight collaborative network traffic measurement system for a data center network in an embodiment of the invention.

[0052] Figure 4 It is a schematic diagram of the results of measuring the load ratio at the maximum flow level of 10 algorithms in the embodiment of the invention under the Fat Tree + CAIDA experimental environment.

[0053] Figure 5 It is a schematic diagram of the results of measuring the load ratio at the maximum flow level of 10 algorithms in the Fat Tree + IMC experimental environment in the embodiment of the invention.

[0054] Figure 6It is a schematic diagram of the results of measuring the load ratio at the maximum flow level of 10 algorithms in the Spine Leaf + CAIDA experimental environment in the embodiments of the present invention.

[0055] Figure 7 It is a schematic diagram of the results of measuring the load ratio at the maximum flow level of 10 algorithms in the Spine Leaf +IMC experimental environment in the embodiments of the present invention.

[0056] Figure 8 It is a schematic diagram of the results of measuring the load ratio at the maximum packet level of 10 algorithms in the Fat Tree + CAIDA experimental environment in the embodiments of the present invention.

[0057] Fig. 9 It is a schematic diagram of the results of measuring the load ratio at the maximum packet level of 10 algorithms in the Fat Tree +IMC experimental environment in the embodiment of the invention.

[0058] Fig.10 It is a schematic diagram of the results of maximum packet level load ratio measurement of 10 algorithms in the Spine Leaf + CAIDA experimental environment in the embodiments of the present invention.

[0059] Fig.11 It is a schematic diagram of the results of measuring the load ratio at the maximum packet level of 10 algorithms in the Spine Leaf +IMC experimental environment in the embodiments of the present invention.

[0060] Fig.12 It is a schematic diagram of the average throughput of switches of 10 algorithms in the Fat Tree +IMC experimental environment in the embodiments of the present invention.

[0061] Fig.13 It is a schematic diagram of the average throughput of switches of 10 algorithms in the Spine Leaf +IMC experimental environment in the embodiments of the present invention.

[0062] Fig.14 It is a schematic diagram of the average absolute error of each flow size measured by 10 algorithms in the Fat Tree + CAIDA experimental environment in the embodiments of the present invention, and the result changes with the memory size.

[0063] Fig.15 It is a schematic diagram of the average absolute error of each flow size measured by 10 algorithms in the Fat Tree +IMC experimental environment in the embodiments of the present invention, and the result changes with the memory size.

[0064] Fig.16It is a schematic diagram of the results of the F1 scores of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree + CAIDA experimental environment, which vary with the memory size.

[0065] Fig.17 It is a schematic diagram of the results of the F1 scores of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree +IMC experimental environment, which vary with the memory size.

[0066] Fig.18 It is a schematic diagram of the results of the average absolute error of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree + CAIDA experimental environment, which varies with the memory size.

[0067] Fig.19 This is a schematic diagram of the results of the average absolute error of the first network flow detection of 10 algorithms in the Fat Tree +IMC experimental environment in the embodiments of the present invention, which varies with the memory size.

[0068] Fig. 20 It is a schematic diagram of the results of the F1 scores of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree + CAIDA experimental environment, and the changes of the first network flow detection threshold.

[0069] Fig.21 It is a schematic diagram of the results of the F1 scores of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree +IMC experimental environment, and the changes of the first network flow detection threshold.

[0070] Fig. 22 It is a schematic diagram of the results of the average absolute error of the first network flow detection of 10 algorithms in the embodiment of the invention under the Fat Tree + CAIDA experimental environment, and the change of the first network flow detection threshold.

[0071] Fig.23 It is a schematic diagram of the results of the average absolute error of the first network flow detection of 10 algorithms in the Fat Tree +IMC experimental environment in the embodiments of the present invention, which changes with the first network flow detection threshold.

[0072] Fig.24 It is a schematic diagram of the structure of an electronic device in an embodiment of the invention. DETAILED DESCRIPTION

[0073] The embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0074] As the core hub of network services, data centers bear the important responsibility of connecting users and services, and are always carrying "massive" network traffic. Due to the need to provide real-time network services to a large number of users at the same time, data center traffic often has the important characteristic of "high speed".

[0075] There are a large number of measurement methods based on the sketch data structure that can efficiently record traffic information, but these methods are limited to single-point measurement and difficult to expand to the whole network traffic measurement. For large data center networks, the huge amount of routing information, link bandwidth limitations, and the resource limitations of the switch itself will greatly affect the strategy formulation and actual deployment of the whole network traffic measurement. In addition, the centralized controller under the software defined network (Software Defined Network, SDN) architecture has a global perspective of the entire network, and the SDN architecture can be used to uniformly manage and control network traffic. However, in the actual measurement process, since the SDN architecture needs to monitor the status of the switches in the network in real time, and frequent collaborative communication between the controller and the switch is required, the architecture itself will bring huge bandwidth consumption to the entire link, which is not an efficient solution.

[0076] Considering that the relevant strategies are formulated based on the topology of the entire network, historical traffic data, and auxiliary information such as routing tables, the purpose is to ensure that any flow entering the network is measured only by one or more fixed measurement nodes. However, the existing collaborative measurement schemes do not effectively solve the problem of duplicate data packet recording, and do not take into account that the real-world traffic distribution is extremely uneven, that is, most of them are small flows, and very few are large flows.

[0077] Taking the real network traffic data set CAIDA19 as an example, if we sort the network flows from large to small, we will find that the "large flows" in the top 2% account for 80% of the total network traffic, while the "small flows" in the bottom 85% only account for about 10% of the total network traffic, indicating that compared with "small flows", the number of "large flows" is smaller, but the number of data packets in each "large flow" is larger. Therefore, although the existing collaborative measurement scheme can make the number of flows measured by each switch close to the same, the actual measurement load of each switch may vary greatly in a short period of time.

[0078] To this end, the present invention provides an embodiment of a lightweight collaborative network traffic measurement method and system for a data center network.

[0079] Please refer to Figure 1 The steps shown in the figure, the lightweight collaborative network traffic measurement method for data center network provided by the embodiment of the present invention includes:

[0080] Step S101, determining that the network flow to be tested is the first network flow or the second network flow, wherein the first network flow is greater than the second network flow;

[0081] Step S102, when the network flow to be measured is a first network flow, evenly distribute the measurement task of the network flow to be measured to each switch in a first path, wherein the first path is a transmission path of the network flow to be measured;

[0082] In the case where the network flow to be measured is the second network flow, the measurement task of the network flow to be measured is allocated to a preset switch, wherein the preset switch is any switch in the second path, and the second path is the transmission path of the network flow to be measured;

[0083] Step S103, controlling the switch to measure and record the flow information of the network flow to be measured according to the assigned measurement task;

[0084] Step S104, responding to the query instruction, obtaining flow information, and calculating the number of data packets in the network flow to be tested.

[0085] Those skilled in the art should understand that a network flow refers to a collection of a series of data packets from a source point to one or more destinations in a network, and the size of a network flow refers to the number of data packets in the network flow; network traffic refers to the sum of all data packets transmitted in the network, regardless of which network flow these data packets belong to.

[0086] The "network flow to be tested" in the embodiments of the present invention can be used to calculate the network traffic carried by the data center within a predetermined measurement period, and to measure the number of data packets in each network flow in the network traffic, that is, each network flow carried by the data center within the predetermined measurement period is the network flow to be tested; it can also be used to measure the size of one or more specified network flows in the time period to be tested, and to measure the number of data packets in the above-mentioned one or more specified network flows, that is, the network flow to be tested refers to the above-mentioned one or more specified network flows.

[0087] In other words, the lightweight collaborative network traffic measurement method for data center networks provided by the embodiments of the present invention can be used to measure the overall network traffic of the data center network, and can also be used to measure the local network traffic of the data center network.

[0088] In the case where the network flow to be tested is the first network flow, evenly distributing the measurement task of the network flow to be tested to each switch in the first path means evenly distributing the data packets in the network flow to be tested to each switch in the first path, because the number of data packets contained in the network flow to be tested in this case is large, and if only one switch is used for measurement, the processing efficiency will be reduced, and the load of the switch will also be too large. At the same time, the network flow to be tested in this case is completely transmitted to each switch in the first path, but each switch only measures the corresponding part of the network flow to be tested, that is, "even distribution" does not mean splitting the network flow to be tested into various parts and then transmitting them to the corresponding switches separately, nor is it limited to average distribution.

[0089] When the network flow to be tested is the second network flow, the measurement task of the network flow to be tested is allocated to a preset switch, and the preset switch is any switch in the second path, which means that the network flow to be tested in this case is measured by only one switch. Because the number of data packets contained in the network flow to be tested in this case is small, the measurement task can be allocated in units of flow.

[0090] In response to the query instruction, the controller may respond to a manual query instruction and then query the flow information recorded in each switch; or the controller may automatically query the flow information recorded in each switch after a preset time arrives according to a built-in program.

[0091] The lightweight collaborative network traffic measurement method for data center networks provided by the embodiments of the present invention determines that the network flow to be measured is the first network flow or the second network flow. When the network flow to be measured is the first network flow, the measurement task of the network flow to be measured is evenly distributed to each switch on its transmission path to achieve packet-level measurement load balancing; when the network flow to be measured is the second network flow, the measurement task of the network flow to be measured is distributed to any switch on its transmission path to achieve flow-level measurement load balancing; thereby avoiding the data packets of each network flow to be measured being repeatedly measured by different switches, and keeping the number of data packets that each switch needs to process similar, and the number of network flows to be measured that each switch needs to process similar, so as to ensure that the actual measurement load of each switch is relatively small within the predetermined time, thereby improving the data packet processing efficiency and the overall measurement accuracy.

[0092] Specifically, step S101, determining whether the network flow to be tested is the first network flow or the second network flow, includes:

[0093] The detector based on the ingress switch calculates the hash value of the data packet in the network flow to be tested and stores the hash value in the corresponding bucket;

[0094] When the same data packet appears in the network flow to be tested, control the counter in the bucket corresponding to the first hash value to increase by 1, the first hash value is the same hash value of the same data packet, wherein the identification field of the same data packet is the same;

[0095] When the hash values ​​of different data packets in the network flow to be tested are the same, a hash conflict occurs, and a counter in the bucket corresponding to the second hash value is controlled to be reduced by 1, and the second hash value is the same hash value of different data packets, wherein the identification fields of different data packets are different;

[0096] When the value of any counter is 0, the network flow to be tested is determined to be the first network flow; when the value of no counter is 0, the network flow to be tested is determined to be the second network flow.

[0097] Exemplarily, the above identification field is a flow label.

[0098] Preferably, before the detector based on the ingress switch calculates the hash value of the data packet in the network flow to be tested, the data packet in the network flow to be tested is sampled according to a preset probability.

[0099] Exemplarily, the above preset probability is less than 0.01, for example, the data packets in the network flow to be tested are sampled with a preset probability of 0.001 or 0.005.

[0100] The detector based on the ingress switch calculates the hash value of the sampled data packet, stores the hash value of the sampled data packet in the corresponding bucket, and then determines whether the network flow to be tested is the first network flow or the second network flow according to the above method.

[0101] Among them, the sampled data packets are used to replace all the data packets in the network flow to be tested. For example, only 0.1% or 0.5% of the data packets in the network flow to be tested need to be detected. This can significantly shorten the time for determining whether the network flow to be tested is the first network flow or the second network flow, and most of the network flows to be tested that belong to the second network flow can be identified and determined in a negligible amount of time.

[0102] Further, after determining that the network flow to be measured is the first network flow, the network flow measurement method further includes:

[0103] The hash value of the network flow to be tested is stored in a target bucket, where the target bucket is a bucket with a counter value of 0;

[0104] Reset the target bucket's counter to 1.

[0105] Exemplarily, the following Algorithm 1 is the content of the algorithm module in the detector:

[0106] Insertion() of LFD

[0107] Input: Packet

[0108] if then / / with preset probability Packet collection sampling

[0109] Extract from / / For the sampled data packets Flow labeling extract

[0110] Calculate the fingerprint and locate the bucket / / Calculate the flow label Hash value And locate the hash value The corresponding bucket

[0111] if then / / in the bucket If empty

[0112] / / Hash value Insert barrel

[0113] / / bucket The counter in increments by 1

[0114] else / / in the bucket When a hash value is inserted

[0115] if then / / Bucket Hash value inserted

[0116] / / bucket The counter in increments by 1

[0117] else / / Bucket Another hash value has been inserted

[0118] / / bucket The counter in decrements by 1

[0119] if then / / in the bucket When the counter in is 0

[0120] / / The bucket The counter in is set to 1

[0121] / / The bucket The hash value inserted in is set to

[0122] LFD is the abbreviation of Large Flow Detector, which is used to detect the first network flow. indivual bucket, , is a hash function, Represents a bucket The hash value in Represents a bucket The counter in .

[0123] Specifically, in step S102, when the network flow to be measured is the second network flow, a hash interval matching requirement is determined based on a preset network topology structure, and a measurement task of the network flow to be measured is allocated to a preset switch according to the hash interval matching requirement and the network topology structure.

[0124] For example, using -tier Clos series network topologies as preset network topologies, such as Fat Tree and Spine Leaf.

[0125] Specifically, the network topology structure Paths of different lengths are denoted as ,path One or more paths The length is set to , h is the total number of layers of switches, path The hash interval that needs to be maintained is recorded as , hash interval The matching requirements are:

[0126] ;

[0127] in, , j represents the j-th layer switch in the h-layer switches; , k represents the kth layer switch in the h layer switches.

[0128] Preferably, the hash interval matching requirement is determined based on the network flow quantity threshold C and a preset network topology structure, and the network flow quantity threshold C needs to satisfy:

[0129] ;

[0130] in, For the Path through the layer switch The number of Hash interval It is understandable that in actual operation, it is not necessary to find the specific value of the network quantity threshold C, but the network quantity threshold C is used as a medium to finally determine the hash interval. Length , thereby balancing the number of network flows that need to be measured on each switch and achieving flow-level measurement load balancing.

[0131] Specifically, in step S103, controlling the switch to record the flow information of the allocated data packet includes:

[0132] Please refer to Figure 2 As shown, the traffic information is recorded separately based on a two-layer collaboration sketch (TC Sketch), wherein the two-layer collaboration sketch is configured on each switch including the ingress switch, and the two-layer collaboration sketch includes two layers of buckets, the first layer bucket B L Including from arrive of Buckets, second layer bucket B S Including from arrive of Buckets, Less than ; First layer bucket B L Each bucket in has d L Cell, second layer bucket B S Each bucket in has d S Cell, in Figure 2 Bucket on the right and barrel Take as an example to illustrate, wherein each cell Cell has a fingerprint and a counter, the fingerprint is used to store the hash value, and the counter is used to count.

[0133] When the network flow to be tested is the first network flow, based on the first layer bucket B L Record the traffic information of the data packets in the network flow to be tested; when the network flow to be tested is the second network flow, based on the second layer bucket BS Record the flow information of data packets in the network flow to be tested.

[0134] Exemplarily, the following Algorithm 2 is the algorithm content of TC Sketch in each switch:

[0135] Insertion() of TC Sketch

[0136] Input: Packet

[0137] Extract and from / / From the data packet Extract flow labels from and element tags And the flag

[0138] Get the length of the path / / Get the path Length

[0139] Get the flow-level interval / / Get the flow level matching interval

[0140] Get the packet-level interval / / Get the package level matching interval

[0141] if then / / In the data packet Used in case of second network flow

[0142] Calculate the value

[0143] if then

[0144] if

[0145] then

[0146] Insert into

[0147] else

[0148] Insert into

[0149] else / / In the data packet Used when it belongs to the first network flow

[0150] Calculate the value

[0151] if then

[0152] Insert into

[0153] Exemplarily, in step S104, the method for calculating the number of data packets in the network flow to be tested is shown in the following Algorithm 3:

[0154] Query() of TC Sketch

[0155] Input: Flow

[0156] Calculate the value

[0157] / / The initial stream size is 0

[0158] for do / / Traverse the path of the switch

[0159] Get the flow-level interval

[0160] if then / / add the interval Recorded value

[0161] is the size queried in

[0162]

[0163] is the size queried in / / Add the interval Recorded value

[0164]

[0165] Return / / Return the final stream size

[0166] Specifically, before the measurement starts, the measurement nodes are determined according to the network topology, including switches for measuring network flows, and corresponding algorithm modules are deployed on these switches.

[0167] Before each measurement cycle begins, the network administrator needs to set the basic parameters of the measurement, such as the flow label or element label of the measurement object, or other traffic information that can characterize the characteristics of the network flow to be measured; since this solution uses a hash function, it is also necessary to calculate the interval that needs to be maintained on the switch based on the current network topology.

[0168] In each measurement cycle, when the ingress switch detects the arrival of a data packet, it uses Algorithm 1 to determine whether the network flow to which the data packet belongs is the first network flow or the second network flow.

[0169] When the network flow to which the data packet belongs is the first network flow, the data packet is marked, and each switch including the ingress switch in the first path measures and records the received data packet according to Algorithm 2.

[0170] When the network flow to which the data packet belongs is the second network flow, the preset switch in the second path measures and records the received data packet according to Algorithm 2.

[0171] After each measurement cycle, the administrator can collect traffic information from the TC Sketch in each of the above switches and calculate the size of any network flow according to Algorithm 3.

[0172] When it is necessary to measure the network traffic carried by the data center within a period of time, the size of each network flow within the period of time is calculated separately, and the network traffic carried by the data center within the period of time can be obtained by adding up the sizes of each network flow.

[0173] Please refer to Figure 3 As shown, based on the above-mentioned lightweight collaborative network traffic measurement method for data center network provided by the embodiment of the invention, the embodiment of the invention also provides a lightweight collaborative network traffic measurement system for data center network, including:

[0174] The server sends or receives the network flow to be tested;

[0175] The switch forwards the network flow to be tested and processes the network flow to be tested;

[0176] The controller determines that the network flow to be tested is the first network flow or the second network flow, wherein the first network flow is greater than the second network flow;

[0177] The controller further evenly distributes the measurement task of the network flow to be measured to each switch in the first path when the network flow to be measured is the first network flow, wherein the first path is the transmission path of the network flow to be measured; and distributes the measurement task of the network flow to be measured to a preset switch when the network flow to be measured is the second network flow, wherein the preset switch is any switch in the second path, and the second path is the transmission path of the network flow to be measured;

[0178] The controller also controls the switch to measure and record the flow information of the network flow to be measured according to the assigned measurement task;

[0179] The controller,also responds to query instructions, obtains traffic information, and calculates the number of,data packets in the network flow to be tested.

[0180] The lightweight collaborative network traffic measurement system for data center networks provided by the embodiments of the present invention has the same beneficial effects as the above-mentioned lightweight collaborative network traffic measurement method for data center networks, which will not be repeated here.

[0181] Exemplarily, the present invention also provides experimental data of the above solution to verify its technical effect, as follows:

[0182] The real-world traffic data sets used for the experiment come from CAIDA and IMC Data Center, respectively. The source IP address in the data set is used as the flow label and the destination IP address is used as the element label. At the same time, two network topologies, Fat Tree and Spine Leaf, are used respectively.

[0183] Therefore, there are four experimental environments in the form of "network topology + traffic data collection":

[0184] Fat Tree +CAIDA, Fat Tree +IMC, Spine Leaf +CAIDA, Spine Leaf +IMC.

[0185] In the above experimental environment, the present invention creates an embodiment that conducts a comparative experiment among 10 algorithms. The hash functions involved in the 10 algorithms are all derived from MURMUR3. The 10 algorithms are:

[0186] LTCM, CMAX, UWRA, DS, WS, WS-NSPA, WS-FCM, CS, CS-NSPA, CS-FCM.

[0187] Among them, LTCM adopts the measurement method provided by the embodiment of the invention. LTCM is the abbreviation of lightweight two-level collaborative measurement, indicating that the measurement method provided by the embodiment of the invention is a lightweight two-level collaborative network traffic measurement method.

[0188] Specifically, FCM is the acronym for flow-level collaborative measurement framework, which is a general flow-level collaborative measurement framework; NSPA is the acronym for Network-based Sketching and Probing for Accurate flow measurement, which is also a measurement framework.

[0189] The two existing single-point measurement algorithms, CS (Chain Sketch) and WS (Waving Sketch), are integrated for FCM and NSPA respectively, and four measurement algorithms, namely WS-NSPA, WS-FCM, CS-NSPA and CS-FCM, are obtained.

[0190] Specifically, CMAX, UWRA, and DS are the acronyms of CountMax (a lightweight collaborative sketch algorithm), Unit Weight Reduction Algorithm, and Distributed Sketch, respectively, all of which are existing measurement algorithms, and the embodiments of the present invention will not be introduced in detail here.

[0191] Please refer to Figures 4 to 7 As shown, the embodiments of the present invention measure the load ratio of the maximum flow level of each of the 10 algorithms, and conduct comparative experiments in four experimental environments: Fat Tree +CAIDA, Fat Tree +IMC, Spine Leaf +CAIDA, and Spine Leaf +IMC.

[0192] Specifically, the maximum flow level measurement load ratio represents the ratio of the maximum number of network flows measured by each switch in the 10 algorithms to the maximum number of network flows measured by each switch in the LTCM algorithm; Figures 4 to 7The maximum flow level measurement load ratio results of the LTCM algorithm are all 1, and the maximum flow level measurement load ratio results of the four algorithms WS-NSPA, WS-FCM, CS-NSPA, and CS-FCM are relatively close to LTCM, but are not lower than 1. Therefore, the maximum flow level measurement load of the LTCM algorithm is the smallest, indicating that the measurement method provided by the embodiment of the present invention has a high measurement accuracy.

[0193] Please refer to Figures 8 to 11 As shown, the embodiments of the present invention measured the load standard deviation of the maximum packet level of each of the 10 algorithms, and carried out comparative experiments in four experimental environments: Fat Tree +CAIDA, Fat Tree +IMC, Spine Leaf +CAIDA, and Spine Leaf +IMC.

[0194] Specifically, the standard deviation of the maximum packet-level load measurement is on the order of 10 6 The smaller the maximum packet level measurement load standard deviation is, the better the measurement load balancing effect of the algorithm is. The LTCM algorithm is Figures 8 to 11 The maximum packet level measurement load standard deviation is the lowest, indicating that the LTCM algorithm has the best measurement load balancing effect among the 10 algorithms, indicating that the measurement method provided by the embodiment of the present invention can achieve effective measurement load balancing.

[0195] Please refer to Fig.12 and Fig.13 As shown, the embodiments of the present invention conduct comparative experiments on the average switch throughput of 10 algorithms in two experimental environments, Fat Tree +IMC and Spine Leaf +IMC.

[0196] Specifically, the unit of average switch throughput is Mpps, which reflects the number of million-level packets that can be processed by each switch of the 10 algorithms per second. Fig.12 and Fig.13 In the examples, the average throughput of the switches using the LTCM algorithm is relatively high, indicating that the measurement method provided by the embodiments of the present invention has a relatively high data packet processing speed.

[0197] Please refer to Fig.14 and Fig.15 As shown, the present invention creates an embodiment of the invention, and conducts comparative experiments in two experimental environments, Fat Tree +CAIDA and Fat Tree +IMC, on the variation of the mean absolute error of each flow size measured for 10 algorithms with the memory size.

[0198] Specifically, the unit of memory is KB. The per-flow size measurement refers to measuring the size of each network flow. The smaller the mean absolute error of the per-flow size measurement, the more accurate the measured network flow size. Fig.14 and Fig.15 In the embodiment, the average absolute error of each flow size measured by the LTCM algorithm in different memories is the smallest, indicating that the measurement method provided by the embodiment of the present invention has a high measurement accuracy.

[0199] Please refer to Fig.16 and Fig.17 As shown, the present invention creates an embodiment of the invention, and conducts comparative experiments on the variation of the F1 scores of the first network flow detection of 10 algorithms with the memory size in two experimental environments, Fat Tree +CAIDA and Fat Tree +IMC.

[0200] Specifically, the higher the F1 score of the first network flow detection, the more accurately the corresponding algorithm can detect the first network flow. Fig.16 and Fig.17 In the results, the F1 scores of the first network flow detection of the LTCM algorithm under different memories are the highest, indicating that the measurement method provided by the embodiment of the present invention can accurately detect the first network flow.

[0201] Please refer to Fig.18 and Fig.19 As shown, the present invention creates an embodiment of the invention and conducts comparative experiments on the mean absolute error of the first network flow detection of 10 algorithms as a function of memory size in two experimental environments: Fat Tree + CAIDA and Fat Tree + IMC.

[0202] Specifically, the smaller the mean absolute error of the first network flow detection is, the more accurately the corresponding algorithm can detect the first network flow. Fig.18 and Fig.19 In the results, the average absolute error of the first network flow detection by the LTCM algorithm under different memories is the smallest, indicating that the measurement method provided by the embodiment of the present invention can accurately detect the first network flow.

[0203] Please refer to Fig. 20 and Fig.21 As shown, the present invention creates an embodiment of the present invention, and conducts comparative experiments in two experimental environments, Fat Tree + CAIDA and Fat Tree + IMC, on the change of the F1 score of the first network flow detection of 10 algorithms with the first network flow detection threshold.

[0204] Specifically, in Fig. 20 and Fig.21In the results, the F1 scores of the first network flow detection of the LTCM algorithm under different first network flow detection thresholds are the highest, indicating that the measurement method provided by the embodiment of the present invention can accurately detect the first network flow.

[0205] Please refer to Fig. 22 and Fig.23 As shown, the present invention creates an embodiment of the invention and conducts comparative experiments in two experimental environments, Fat Tree + CAIDA and Fat Tree + IMC, on the change of the average absolute error of the first network flow detection of 10 algorithms with the first network flow detection threshold.

[0206] Specifically, in Fig. 22 and Fig.23 In the figure, the average absolute error of the first network flow detection by the LTCM algorithm under different first network flow detection thresholds is the smallest, indicating that the measurement method provided by the embodiment of the present invention can accurately detect the first network flow.

[0207] In summary, the lightweight collaborative network traffic measurement method for data center networks provided by the embodiments of the present invention can achieve both high data packet processing efficiency and high overall measurement accuracy while measuring fine-grained network traffic in large data centers.

[0208] The present invention also provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the present invention.

[0209] The present invention also provides a computer program product, including a computer program, wherein the computer program, when executed by a processor of a computer, is used to enable the computer to execute the above method provided by the present invention.

[0210] The present invention also provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor. The memory stores a computer program executable by the at least one processor, and the computer program is used to enable the electronic device to perform the method provided by the present invention.

[0211] refer to Fig.24, a block diagram of an electronic device that can be used as a server or client of an embodiment of the invention will now be described, which is an example of hardware devices that can be applied to various aspects of the invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or required herein.

[0212] like Fig.24 As shown, the electronic device includes a computing unit 2401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 2402 or a computer program loaded from a storage unit 2408 to a random access memory (RAM) 2403. In the RAM 2403, various programs and data required for the operation of the electronic device can also be stored. The computing unit 2401, the ROM 2402, and the RAM 2403 are connected to each other via a bus 2404. An input / output (I / O) interface 2405 is also connected to the bus 2404.

[0213] Multiple components in the electronic device are connected to the I / O interface 2405, including: an input unit 2406, an output unit 2407, a storage unit 2408, and a communication unit 2409. The input unit 2406 can be any type of device that can input information to the electronic device, and the input unit 2406 can receive input digital or character information, and generate key signal input related to user settings and / or function control of the electronic device. The output unit 2407 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 2408 can include but is not limited to a disk, an optical disk. The communication unit 2409 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, and / or a wireless communication transceiver, such as a Bluetooth device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0214] The computing unit 2401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 2401 include, but are not limited to, a CPU, a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 2401 performs the various methods and processes described above. For example, in some embodiments, the method embodiments created by the present invention may be implemented as a computer program, which is tangibly contained in a machine-readable medium, such as a storage unit 2408. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via ROM 2402 and / or a communication unit 2409. In some embodiments, the computing unit 2401 may be configured to perform the above-described method in any other appropriate manner (e.g., by means of firmware).

[0215] The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the computer program, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0216] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, or infrared systems, devices, or equipment, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0217] It should be noted that the term "including" and its variations used in the embodiments of the present invention are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present invention are illustrative and not restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more". The descriptions of the terms "first", "second", etc. are for descriptive purposes only and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated.

[0218] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0219] The various steps described in the method implementation methods provided by the embodiments of the present invention can be performed in different orders and / or in parallel. In addition, the method implementation methods may include additional steps and / or omit the steps shown. The scope of protection of the present invention is not limited in this respect.

[0220] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the invention. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same and similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0221] The above-described embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of protection. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the attached claims.

Claims

1. A lightweight collaborative network traffic measurement method for data center networks, characterized in that: include: Determine that the network flow to be tested is a first network flow or a second network flow, wherein the first network flow is greater than the second network flow; In a case where the network flow to be measured is the first network flow, evenly distributing the measurement task of the network flow to be measured to each switch in a first path, wherein the first path is a transmission path of the network flow to be measured; In the case where the network flow to be measured is the second network flow, allocating the measurement task of the network flow to be measured to a preset switch, wherein the preset switch is any switch in the second path, and the second path is the transmission path of the network flow to be measured; Controlling the switch to measure and record the flow information of the network flow to be measured according to the assigned measurement task; In response to the query instruction, the flow information is obtained, and the number of data packets in the network flow to be tested is calculated.

2. The lightweight collaborative network traffic measurement method for data center networks according to claim 1 is characterized in that: Determining whether the network flow to be tested is the first network flow or the second network flow includes: The detector based on the ingress switch calculates the hash value of the data packet in the network flow to be tested, and stores the hash value in the corresponding bucket; When the same data packet appears in the network flow to be tested, control the counter in the bucket corresponding to the first hash value to increase by 1, wherein the first hash value is the same hash value of the same data packet, wherein the identification fields of the same data packets are the same; When the hash values ​​of different data packets in the network flow to be tested are the same, a hash conflict occurs, and a counter in a bucket corresponding to a second hash value is controlled to be reduced by 1, wherein the second hash value is the same hash value of the different data packets, wherein the identification fields of the different data packets are different; Wherein, when the value of any of the counters is 0, the network flow to be tested is determined to be the first network flow; when the value of none of the counters is 0, the network flow to be tested is determined to be the second network flow.

3. The lightweight collaborative network traffic measurement method for data center networks according to claim 2 is characterized in that: After determining that the network flow to be tested is the first network flow, the method further includes: Storing the hash value of the network flow to be tested into a target bucket, wherein the target bucket is a bucket whose counter value is 0; Reset the value of the counter of the target bucket to 1.

4. The lightweight collaborative network traffic measurement method for data center networks according to claim 2 is characterized in that: Determining whether the network flow to be tested is the first network flow or the second network flow also includes: Before calculating the hash value of the data packet in the network flow to be tested, the detector based on the ingress switch samples the data packet in the network flow to be tested according to a preset probability; The detector based on the ingress switch calculates the hash value of the sampled data packet, and stores the hash value of the sampled data packet in a corresponding bucket.

5. The lightweight collaborative network traffic measurement method for data center networks according to claim 1 is characterized in that: The method further comprises: In a case where the network flow to be tested is the second network flow, determining a hash interval matching requirement based on a preset network topology structure; Wherein, the first Paths of different lengths are denoted as , the path is one or more, the path The length is set to , h is the total number of layers of the switch, the path The hash interval that needs to be maintained is denoted as , the hash interval The matching requirements are: ； in, , j represents the j-th layer switch in the h-layer switches; , k represents the k-th layer switch in the h-layer switches; According to the hash interval matching requirement and the network topology structure, the measurement task of the network flow to be measured is allocated to a preset switch.

6. The lightweight collaborative network traffic measurement method for data center networks according to claim 5 is characterized in that: Determine hash interval matching requirements based on the preset network topology, including: The hash interval matching requirement is determined based on the network flow quantity threshold C and the preset network topology structure. The network flow quantity threshold C needs to meet the following requirements: ； in, For the The path through the layer switch The number of For the hash interval Length; According to the network flow quantity threshold C, the hash interval matching requirement and the network topology structure, the measurement task of the network flow to be measured is allocated to the preset switch.

7. The lightweight collaborative network traffic measurement method for data center networks according to claim 1 is characterized in that: Controlling the switch to record the flow information of the allocated data packet, including: The flow information is recorded separately based on a two-layer collaboration sketch, wherein the two-layer collaboration sketch is configured on each of the switches, and the two-layer collaboration sketch includes two layers of buckets, a first layer bucket B L include Buckets, second layer bucket B S include Buckets, Less than ; When the network flow to be tested is the first network flow, based on the first layer bucket B L Record the flow information of the data packets in the network flow to be tested; if the network flow to be tested is the second network flow, based on the second layer bucket B S The flow information of the data packets in the network flow to be tested is recorded.

8. The lightweight collaborative network traffic measurement method for data center networks according to claim 1 is characterized in that: The traffic information is a flow label or an element label.

9. A lightweight collaborative network traffic measurement system for data center networks, characterized in that: include: Server, sends or receives the network flow to be tested; The switch forwards the network flow to be tested and processes the network flow to be tested; The controller determines that the network flow to be tested is a first network flow or a second network flow, wherein the first network flow is greater than the second network flow; The controller, further, when the network flow to be measured is the first network flow, evenly distributes the measurement task of the network flow to be measured to each switch in a first path, wherein the first path is a transmission path of the network flow to be measured; when the network flow to be measured is the second network flow, distributes the measurement task of the network flow to be measured to a preset switch, wherein the preset switch is any switch in a second path, and the second path is a transmission path of the network flow to be measured; The controller further controls the switch to measure and record the flow information of the network flow to be measured according to the measurement task assigned to it; The controller also responds to the query instruction, obtains the flow information, and calculates the number of data packets in the network flow to be tested.

10. An electronic device, comprising: A processor, and a memory storing a program, characterized in that the program includes instructions, which, when executed by the processor, cause the processor to execute a lightweight collaborative network traffic measurement method for a data center network according to any one of claims 1 to 8.

Images