WO2019210948A1 - Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel - Google Patents

Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel Download PDF

Info

Publication number
WO2019210948A1
WO2019210948A1 PCT/EP2018/061227 EP2018061227W WO2019210948A1 WO 2019210948 A1 WO2019210948 A1 WO 2019210948A1 EP 2018061227 W EP2018061227 W EP 2018061227W WO 2019210948 A1 WO2019210948 A1 WO 2019210948A1
Authority
WO
WIPO (PCT)
Prior art keywords
ble
flow
flows
controller
switch
Prior art date
Application number
PCT/EP2018/061227
Other languages
English (en)
Inventor
Stefano PARIS
Jeremie Leguay
Antonio Capone
Ilario FILIPPINI
Davide SANVITO
Original Assignee
Huawei Technologies Co., Ltd.
Politecnico Di Milano - Dipartimento Di Elettronica, Informazione E Bioingegneria
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 Huawei Technologies Co., Ltd., Politecnico Di Milano - Dipartimento Di Elettronica, Informazione E Bioingegneria filed Critical Huawei Technologies Co., Ltd.
Priority to PCT/EP2018/061227 priority Critical patent/WO2019210948A1/fr
Publication of WO2019210948A1 publication Critical patent/WO2019210948A1/fr

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/10—Flow control; Congestion control
    • H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
    • H04L47/2441—Traffic characterised by specific attributes, e.g. priority or QoS relying on flow classification, e.g. using integrated services [IntServ]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/70—Admission control; Resource allocation
    • H04L47/82—Miscellaneous aspects
    • H04L47/822—Collecting or measuring resource availability data
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/64—Routing or path finding of packets in data switching networks using an overlay routing layer
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L47/00—Traffic control in data switching networks
    • H04L47/70—Admission control; Resource allocation
    • H04L47/82—Miscellaneous aspects
    • H04L47/829—Topology based

Definitions

  • the present invention relates to the technical field of Traffic Engineering (TE), specifically of flows in a Software Defined Network (SDN).
  • SDN Software Defined Network
  • the present invention presents to this end a controller for the SDN and a switch for the SDN, respectively, which can be used to determine and control so-called Bottleneck-Limited Elastic (BLE) flows in the SDN.
  • BLE Bottleneck-Limited Elastic
  • the present invention relates also to a corresponding control method.
  • TE plays a crucial role in any network management, since it permits optimizing and load balancing the utilization of network resources.
  • Sophisticated algorithms have been designed to implement efficient TE.
  • SDN controllers integrate TE methods to continuously optimize the SDN.
  • An important challenge for TE is to be able to react when a congestion occurs, in order to optimally allocate the available capacity of network links.
  • NUM Network Utility Maximization
  • the present invention aims to improve the conventional solutions.
  • the objective problem addressed by the present invention is to detect elastic flows, whose rate/throughput is limited by congestion (i.e. BLE flows) inside the network. Especially, this should be achieved with a minimum signaling overhead. Further, the present invention aims at providing a suitable reaction, if BLE flows are detected. In particular, new flow rules should be installed into switches, in order to e.g. assign more resources to these BLE flows and thus let them grow.
  • the present invention proposes a switch, which is able to detect elastic flows that are potentially limited by network congestion (BLE flows). Further, it proposes a controller, which is able - in collaboration with the switch - to decide, which flows are actually limited in terms of throughput. Further, the invention proposes a way to allow the expansion of these elastic flows, for example, by the controller allocating more network resources or different paths to the determined BLE flows. For instance, BLE flows may be transferred to paths where they can grow, so as to improve completion time of network intensive tasks, e.g. in datacenters.
  • the present invention uses a combination of local detection (at the switch) and a global detection (at the controller) of BLE flows, in order to reach high detection accuracy at a low overhead cost.
  • a flow is identified as elastic, when its rate is limited on a bottleneck link.
  • a bottleneck link for a given data flow is a link that is saturated and of all the flows sharing this link, the given data flow achieves maximum data rate network wide.
  • a link is saturated if the rate of the aggregated flow traversing it is larger than a certain threshold, which can be set to any percentage of the overall capacity.
  • a first aspect of the present invention provides a controller for a SDN, the controller being configured to receive information regarding one or more BLE flow candidates from a switch, analyze the BLE flow candidates over all links of their routing paths to determine one or more BLE flows, and modify resources of the determined one or more BLE flows.
  • the controller of the first aspect receives local detection results from the switch, namely the information regarding the BLE flow candidates. It is then able to use its knowledge of the links and routing paths of these flows to perform a global detection for detecting BLE flows. Accordingly, the controller is able to detect BLE flows with a higher accuracy and a lower overhead cost. The controller is further able to support growth of the determined BLE flows, namely by modifying their resources.
  • the controller is configured to analyze the BLE flow candidates further based on a network topology and/or one or more link capacities on their routing paths to determine the one or more BLE flows.
  • the controller is configured to analyze, for each BLE flow candidate, an evolution of its flow rate over all links of its routing path to determine whether the BLE flow candidate is a BLE flow or not.
  • the controller is configured to analyze, for each BLE flow candidate, side information, particularly a type of the flow and/or a type of an application generating the flow, determine that the BLE flow candidate is a BLE flow according to the side information.
  • the side information further supports the determination of the BLE flows carried out by the controller, and thus leads to even more accurate results.
  • the controller is configured to, in order to modify the resources of the one or more determined BLE flows, allocate additional resources to the determined one or more BLE flows, particularly based on the network topology, link capacities on their routing paths, flow rate, and/or side information.
  • the determined BLE flows are allowed to grow, without resulting in further congestion in the network.
  • the controller is configured to, in order to modify the resources of the one or more determined BLE flows, compute a new network configuration.
  • the new network configuration particularly optimizes the paths of the determined BLE flows, in order to increase their throughput.
  • the controller is configured to update at least one switch according to the new network configuration.
  • the one or more switches are made aware of the new network configuration, and can support its implementation for an increased network throughput.
  • the controller is configured to, in order to modify the resources of the determined one or more BLE flows, by at least one of: rerouting a path of the one or more BLE flows, moving the one or more BLE flows to less congested paths, providing the one or more BLE flows with more bandwidth, providing the one or more BLE flows with an additional routing path for multi-paths routing, or notify another entity to modify the resources of the one or more BLE flows.
  • the controller is configured to maintain a flow table including for each flow in the network at least one of: a flow identifier, a flow volume, an elastic flow flag indicating whether the flow is an elastic flow or not.
  • the controller is able to monitor all the flows in the network, and thus can implement a more efficient detection of and resource allocation to BLE flows.
  • a second aspect of the present invention provides a switch for a SDN, configured to determine one or more Bottleneck-Limited Elastic, BLE, flow candidates from flows traversing the switch, and send information regarding the BLE flow candidates to a controller.
  • BLE Bottleneck-Limited Elastic
  • the switch is thus able to carry out a local detection of BLE flows, and provide the result as the information about the BLE flow candidates to the controller.
  • the controller can more efficiently and more accurately use its global view and knowledge of the network, in order to determine the BLE flows.
  • the switch is configured to determine the one or more BLE flow candidates by evaluating a bandwidth of one or more outgoing links or ports of the switch consumed by the flows traversing the switch.
  • the switch is able to carry out an even more accurate detection of the candidates.
  • the switch is configured to determine the one or more BLE flow candidates by determining flows having a size that is larger than an average flow size to be elephant flows, and identifying all elephant flows traversing a congested link or port of the switch, wherein a link or port is congested, if an overall rate of transmitted flows matches a maximum transmission rate of the link or port.
  • the switch is able to more efficiently carry out an accurate detection of the BLE flow candidates.
  • the switch is further configured to send an identifier of the congested link or port and identifiers of the identified elephant flows as the information regarding the BLE flow candidates to the controller.
  • This information leads to only very low signaling overhead, but allows the controller to identify the flows determined as BLE flow candidates easily.
  • the switch is further configured to maintain a flow context table including for each flow traversing the switch at least one of: a flow identifier, a total number of bytes received from the flow, an elephant flow flag indicating whether the flow is an elephant flow or not.
  • the switch is further configured to maintain a list of congested links or ports of the switch.
  • the switch can more quickly determine, which flows potentially are BLE flows (i.e. BLE flow candidates).
  • a third aspect of the present invention provides a method for controlling BLE flows in a SDN, the method comprising determining one or more BLE flow candidates from flows traversing one or more switches, and analyzing the BLE flow candidates over all links of their routing paths to determine one or more BLE flows, and modifying resources of the determined one or more BLE flows.
  • a fourth aspect of the present invention provides a computer program product storing a program code for controlling a controller according to the first aspect or any of its implementation forms and/or a switch according to the second aspect or any of its implementation forms, or for performing, when implemented on a computer, the method according to the third aspect.
  • FIG. 1 shows a controller according to an embodiment of the present invention.
  • FIG. 2 shows a switch according to an embodiment of the present invention.
  • FIG. 3 shows an architecture of a SDN controller according to an embodiment of the present invention.
  • FIG. 4 shows an overview of steps for detecting and expanding BLE flows.
  • FIG. 5 shows an architecture of a SDN controller according to an embodiment of the present invention.
  • FIG. 6 shows a general description of steps S1-S3 and C1-C4 for detecting and expending BLE flows at the controller and switch according to embodiments of the present invention.
  • FIG. 7 shows details of step Sl.
  • FIG. 8 shows details of step S2.
  • FIG. 9 shows an example of an algorithm for rerouting a determined BLE flow
  • FIG. 10 shows an example of an algorithm according to a MILP model
  • FIG. 11 shows an example of an algorithm according to a MILP model
  • FIG. 12 shows a method according to an embodiment of the present invention.
  • FIG. 1 shows a controller 100 according to an embodiment of the present invention.
  • the controller 100 is particularly for a SDN, i.e. it is an SDN controller.
  • the controller 100 may be implemented by processing circuitry, e.g. by one or more processors.
  • the controller 100 may also be or be provided by a device, which comprises at least one processor for implementing the SDN controller functions.
  • the controller 100 is configured to receive information 101 regarding one or more BLE flow candidates 102 from at least one switch 110.
  • the switch 110 may have previously determined these BLE flow candidates 102 from flows passing through it (this is explained below with respect to FIG. 2).
  • the controller 100 is configured to analyze the BLE flow candidates 102 over all links of their routing paths (in the SDN) to determine one or more BLE flows 103.
  • the routing path of a flow may pass one or multiple switches, each switch having at least an ingoing and outgoing link and/or port.
  • the controller 100 is configured to modify 104 resources of the determined one or more BLE flows 103, e.g. by allocating additional resources to the determined BLE flows 103.
  • This can be achieved, for instance, by computing a new network configuration, by rerouting a path of one or more BLE flows 103, by moving one or more BLE flows 103 to different paths, particularly to less congested paths, by providing one or more BLE flows 103 with more bandwidth, by providing one or more BLE flows 103 with an additional routing path for multi-paths routing, and/or by notifying another entity to modify the resources of one or more BLE flows 103.
  • FIG. 2 shows a switch 110 according to an embodiment of the present invention.
  • the switch is particularly suited for an SDN, i.e. it is an SDN switch.
  • the switch 110 may be implemented by processing circuitry, e.g. by one or more processors.
  • the switch 110 may also be or be provided by a device, which comprises at least one processor for implementing switch functions.
  • the switch 110 is configured to determine one or more BLE flow candidates 102 from flows 200 traversing the switch 110, and to send information 101 regarding the BLE flow candidates 102 to a controller 100 (which may be the controller of FIG. 1).
  • FIG. 3 shows a general architecture of a SDN controller 100 according to an embodiment of the present invention, which builds on the SDN controller 100 shown in FIG. 1.
  • information collected by a monitoring module 302 like the rate of the flows and the gradient of their throughout, may be used by a BLE detection module 300 to detect and tag elastic flows that are limited by the congestion (BLE flows 103).
  • the BLE detection module 300 may be part of the controller 100 (implemented by the controller 100) or may be distributed over controller 100 and switches 110 (explained later). Further, a BLE flow rerouting module 301 of the controller 100 may be responsible for modifying 104 the resources for the determined BLE flows 103, e.g. by computing a new routing configuration or by assigning more resources to tagged flows.
  • the BLE flow detection module 301 relies on information from both the switches 110 (the information 101) and the controller 100 (information provided from the monitoring module 302), in order to detect the elastic flows that are limited by congestion inside the network.
  • the BLE flow detection module 301 may reallocate network resources to elastic flows, which have been detected by the BLE flow detection module 300. Examples of actions executed by this BLE flow rerouting module 301 are the rerouting of a determined BLE flow 103 on a different path, the reservation of additional bandwidth of the link traversed by the BLE flow 103, the assignment of new paths for end-to-end flows that support path splitting (e.g., Multi-Path TCP), etc.
  • path splitting e.g., Multi-Path TCP
  • FIG. 4 shows three main steps for detection and rerouting of BLE flows 103, which are carried out by one or more switches 110 according to FIG. 2 and at least one controller 100 according to FIG. 1.
  • the detection of the BLE flows 103 is triggered in step 1 by at least one switch 110 that locally detects potential candidate BLE flows 102, particularly by evaluating the bandwidth of outgoing ports consumed by the flows 200 traversing the switch 110. Since the flows 200 can be limited by any device on the end-to-end path, the controller 100 in step 2 installs monitoring rules for the BLE flow candidates 102 suggested by the switch 110. By monitoring the evolution of the flow rates over all links of the end-to-end path, the controller 100 is able to discriminate BLE flows 103 (from false candidates 102).
  • step 3 the SDN controller 100 decides to reroute only the one or more BLE flows 103, whose rate eventually grows.
  • FIG. 5 bases on FIG. 3, and shows again the general architecture of the SDN controller 100 and the module 300 for BLE flow detection (BLE-FD) and the module 301 for BLE Flow Expansion (BLE-FE) or rerouting.
  • the BLE-FD module 301 in FIG. 5 is a distributed module operating both in the switches 110 and the controller 100.
  • a first BLE-FD agent (or a first set of BLE-FD sub-modules) may operate in the switches 110, and in this case detects potential BLE flows 102 that are communicated to the controller 100.
  • a second BLE-FD agent may operate in the controller 100, and in this case receives from the switches 110 the candidate BLE flows 102, and performs further analysis to filter out false positives using the global view of the network status.
  • the BLE-FE module 301 may allocate more network resources to the determined BLE flows 103 (e.g., reroute the BLE flows 103 on different paths, or assign additional paths to the BLE flows 103) so that they can expand. Both modules 300 and 301 may be continuously executed by the controller 100.
  • the BLE-FD module 300 may be executed to update the set of BLE flows 103, whenever new flows appear or old flows disappear from the network.
  • the BLE-FE module 301 may be executed to keep the network in its best configuration as the system evolves.
  • FIG. 6 shows an example of a distributed BLE-FD module 300, which includes seven BLE-FD sub-modules: three sub-modules operating in a switch 110 (performing steps labelled Sl, S2, and S3) and four sub-modules operating in the controller 100 (performing steps Cl, C2, C3, and C4).
  • the three sub-modules operating in the switch 110 may be continuously executed and detect locally a list of potential BLE flows 102.
  • the two sub-modules operating in the controller 100 may be triggered by the information sent by the switch 110 through the south-bound interface, and filter out false positive BLE flows (i.e., flows that locally appears to be limited by congestion occurring on the outgoing link or port of the switch 110, but they are actually limited for other reasons).
  • the BLE-FD sub-modules operating in the switch 110 are now explained in more detail with respect to the performed steps S1-S3, i.e. functions the switch 110 is configured to carry out.
  • SI The switch 110 samples the flows 200 traversing it, in order to detect elephants, namely flows 200 whose size is larger than the average of flows size.
  • the switch 110 identifies the flows 200 marked as elephants that are crossing a congested link, namely a link where the overall rate of the transmitted traffic is larger or equal than the maximum transmission rate of the link (i.e., port rate). Using the information about the flows that have been marked as elephant and the links that have been marked as congested, the switch 110 identifies the BLE flow candidates 102, i.e. elephant flows whose rate is limited by the congestion experienced on one of its links.
  • the switch 110 sends to the controller 100 the information 101, particularly a message containing the identifier of the congested links and the identifiers of BLE flow candidates 102, namely flows marked as elephants that are crossing a congested link.
  • the BLE-FD sub-modules operating in the controller 100 are now explained in more detail with respect to the performed steps C1-C4, i.e. functions the controller 110 is configured to carry out.
  • the controller 100 receives the information 101, particularly messages from one or more switches 110 indicating the congested links and the identifiers of the flows marked as BLE flow candidates 102.
  • C2 The controller 100 aggregates all information about BLE flow candidates 102 received by the one or more switches 100, and uses side information provided e.g. by the monitoring module 302 to filter out flows that have been misclassified as BLE flows 103.
  • C3 Once the controller 100 has identified the right BLE flows 103, it computes a new network configuration to optimize a pre-defined objective function. The optimization process allocates more resources to BLE flows 103 (e.g., new routing path, more bandwidth, additional routing parts for multi-paths connections), in order to let them expand their rate/throughput.
  • a SDN switch 110 may keep a Flow Context Table (FCT), in which it stores ⁇ Flow ID, F n
  • the SDN switch 110 may keep a list of links marked as congested (Cong_Link).
  • Step Sl flow sampling and memory occupation limitation
  • FIG. 7 may include the following details with respect to FIG. 7 :
  • Sl.l Every interval of T seconds, the packets flowing to every port of the switch 110 are monitored for W consecutive seconds (W ⁇ T), wherein T and W may be received from the controller 100. For each packet n, the packet size L n , the arrival timestamp T n , and the fields needed to assign a flow ID F n (according to the desired flow classification mle) are recorded.
  • step S1.2 If the arrived packet has a flow ID already in the FCT, it is continued with step S1.3 below, otherwise it is continued with step S1.4 below.
  • step S1.5 If flow F n has been inserted into the FCT, go to step S1.6, otherwise go to step S2.
  • Step S2 (detection of BLE flow candidates 102) may include the following details with respect to FIG. 8:
  • step S2.1 If Totn is greater than threshold E received from the controller 100, continue to step S2.3, otherwise continue to step S2.2
  • step S2.4 If measured port rate is greater than the threshold B received from the controller 100, continue to step S2.6, otherwise continue to step S2.5
  • the switch 110 can determine if a flow F n is a BLE flow candidate 102 as follow.
  • step C2.8 continue to step C2.8, otherwise continue to step C2.9
  • the flow F n is marked as BLE flow candidate 102.
  • SB Southbound
  • step Cl the controller 100 receives the records and list Cong_Link sent by one or more switches 110 via the SB interface. It uses the received information (F n and Tot n ) to fill the table FT.
  • step C2 filter BLE flow candidates 102 that are not elastic
  • the controller 100 performs the final BLE detection.
  • the controller 100 particularly based on the side information received from the monitoring module 301, and the information 101 about the BLE flow candidates 102 received by the switches 110 through the SB interfaces, is configured to filter out the flow candidates that have been misclassified as BLE flows (i.e. misclassified as BLE flow candidates 102 by the switch 110). To this end, the controller 110 just uses the side information indicating which flows are not elastic to correct the decision taken by the switches 110.
  • step C3 reroute flows to expand BLE flows 103
  • the controller 100 implements the expansion of determined BLE flows 103 by acting on network flows routing.
  • Flows can be rerouted according to the algorithms described in the following, or something else can be done to improve BLE performance, like opening new connections and operating Multipath Transmission Control Protocol (MPTCP) sub-flows.
  • MPTCP Multipath Transmission Control Protocol
  • a first algorithm Al is now described (referred also to as“Greedy BLE rerouting”) with respect to FIG. 9.
  • A1.4 Compute the shortest path p for flow Fi according to the O-D pair. Different metrics can be used, like number of hops, inverse of link residual capacity, etc.
  • A1.5 If the path p exists, continue to step A1.6, otherwise go to step A1.8
  • A1.8 Remove flow Fi from L and go back to step A1.2 if list L is not empty, otherwise stop.
  • a second algorithm A2 is now described (referred also to as “Optimal network rerouting”).
  • MILP Mixed Integer Linear Programming
  • the result of the model is the routing of BLE flows 103 that maximizes their throughput.
  • Ci j Capacity of link ( i,j )
  • the objective function (1) and constraint (2) provide the maximization of a linear-piecewise approximation of a function that expresses the rate of competing elastic flows according to congestion control parameters and RTTs.
  • the function in Vojnovic et al,“Global Fairness of additive-increase and multiplicative-decrease with heterogeneous round-trip times” Proceedings of IEEE INFOCOM 2000, pp. 1303-1312 can be used.
  • the constraint (3) is a flow-balance constraint for proper flows routing, while the constraint (4) limits the sum of the flow rates along a link not to exceed the link capacity.
  • the constraint (5) states that in order to have a non-zero rate along a link for a flow, the link must belong to the flow path.
  • the constraint (6) forces flows to be routed through a single non-split path.
  • MILP model Another example of an MILP model, which is similar to the one mentioned above, is shown in FIG.11.
  • This MILP model can be used if inelastic flows are allowed to change their paths as well. This allows to even further maximize BLE flows throughput.
  • decision variables for inelastic flows (INEL super/subscripts) similar to those in the previous model, the new formulation is the following:
  • FIG. 12 shows a method 1200 according to an embodiment of the present invention.
  • the method 1200 is in particular for controlling BLE flows 103 in a SDN.
  • the method 1200 comprises a step 1201 of determining one or more BLE flow candidates 102 from flows 200 traversing one or more switches 110. Further, the method 1200 comprises a step 1202 of analyzing the BLE flow candidates 102 over all links of their routing paths to determine one or more BLE flows 103. Further, the method 1200 comprises a step 1203 of modifying 104 resources of the determined one or more BLE flows 103.
  • the method 1200 may be implemented by the controller 100 and the switch 110, respectively.
  • step 1201 may be implemented by the switch 110
  • steps 1202 and 1203 may be implemented by the controller 100.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

La présente invention concerne l'ingénierie de trafic dans un réseau défini par logiciel (SDN). En particulier, l'invention présente un contrôleur SDN et un commutateur configuré pour détecter des flux élastiques limités par étranglement (Bottleneck-Limited Elastic, BLE) dans le réseau. En particulier, le contrôleur est configuré pour recevoir des informations concernant un ou plusieurs flux BLE candidats à partir du commutateur, pour analyser les flux BLE candidats sur toutes les liaisons de leurs trajets d'acheminement pour déterminer un ou plusieurs flux BLE, et pour modifier les ressources du ou des flux BLE déterminés. Le commutateur est configuré pour déterminer un ou plusieurs flux BLE candidats à partir de flux traversant le commutateur, et pour envoyer des informations concernant flux BLE candidats à un contrôleur.
PCT/EP2018/061227 2018-05-02 2018-05-02 Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel WO2019210948A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/EP2018/061227 WO2019210948A1 (fr) 2018-05-02 2018-05-02 Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/EP2018/061227 WO2019210948A1 (fr) 2018-05-02 2018-05-02 Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel

Publications (1)

Publication Number Publication Date
WO2019210948A1 true WO2019210948A1 (fr) 2019-11-07

Family

ID=62116421

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2018/061227 WO2019210948A1 (fr) 2018-05-02 2018-05-02 Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel

Country Status (1)

Country Link
WO (1) WO2019210948A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116248578A (zh) * 2022-12-21 2023-06-09 重庆邮电大学 软件定义网络中综合链路负载与带宽碎片化的流调度方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150163145A1 (en) * 2013-12-09 2015-06-11 Nicira, Inc. Reporting elephant flows to a network controller
WO2017167390A1 (fr) * 2016-03-31 2017-10-05 Huawei Technologies Co., Ltd. Dispositif et procédé d'analyse de type de flux d'informations
WO2018004639A1 (fr) * 2016-07-01 2018-01-04 Hewlett Packard Enterprise Development Lp Équilibrage de charge

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150163145A1 (en) * 2013-12-09 2015-06-11 Nicira, Inc. Reporting elephant flows to a network controller
WO2017167390A1 (fr) * 2016-03-31 2017-10-05 Huawei Technologies Co., Ltd. Dispositif et procédé d'analyse de type de flux d'informations
WO2018004639A1 (fr) * 2016-07-01 2018-01-04 Hewlett Packard Enterprise Development Lp Équilibrage de charge

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
VOJNOVIC ET AL.: "Global Fairness of additive-increase and multiplicative-decrease with heterogeneous round-trip times", PROCEEDINGS OF IEEE INFOCOM, 2000, pages 1303 - 1312, XP001044224

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116248578A (zh) * 2022-12-21 2023-06-09 重庆邮电大学 软件定义网络中综合链路负载与带宽碎片化的流调度方法

Similar Documents

Publication Publication Date Title
CN109039884B (zh) 网络路径预测和选择的方法、设备和介质
EP3047609B1 (fr) Systèmes et procédé pour la réconfiguration de routes
JP5750714B2 (ja) 計算機システム、仮想サーバ配置方法及び配置制御装置
KR102036056B1 (ko) 중앙 제어기들에 의한 네트워크들에서의 지연-기반 트래픽 레이트 제어
US8335162B2 (en) Methods and apparatus to control traffic in a packet-switched network
WO2014068426A1 (fr) Procédé d'équilibrage dynamique de charges de flux de réseau sur des interfaces de retard
JP2013150134A5 (fr)
WO2011131738A1 (fr) Système de sondage de gestion d'encombrement de données de réseau
US11575599B2 (en) Multi-tier deterministic networking
US10341224B2 (en) Layer-3 flow control information routing system
CN108476175B (zh) 使用对偶变量的传送sdn流量工程方法与系统
KR20170113807A (ko) 서비스 기능 체이닝 시스템 및 그 방법
CN A proactive flow admission and re-routing scheme for load balancing and mitigation of congestion propagation in SDN data plane
Abdollahi et al. Flow-aware forwarding in SDN datacenters using a knapsack-PSO-based solution
Abbou et al. Towards SDN-based deterministic networking: Deterministic E2E delay case
KR101608818B1 (ko) 가상 서비스 체인 설정을 위한 네트워크 시스템 및 컴퓨터 프로그램
WO2019210948A1 (fr) Contrôleur et commutateur pour l'ingénierie du trafic dans un réseau défini par logiciel
EP3485618B1 (fr) Dispositif et procédé de gestion de connexions de bout en bout
CN111901237B (zh) 源路由选路方法及系统、相关设备及计算机可读存储介质
EP3468122B1 (fr) Admission de paquets
CN108040018B (zh) 一种网络功能虚拟化下的细粒度网络流调度方法及系统
Gunavathie et al. MLPRS: A Machine Learning-Based Proactive Re-Routing Scheme for flow classification and priority assignment
US20230198895A1 (en) Methods and systems for adaptive stochastic-based load balancing
CN112822233A (zh) 流量重定向方法及装置
Wu et al. Alleviating network congestion for HPC clusters with fat-tree interconnection leveraging software-defined networking

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18722467

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18722467

Country of ref document: EP

Kind code of ref document: A1