WO2018042459A1

WO2018042459A1 - Adaptive and seamless traffic steering among multiple paths based on application qoe needs

Info

Publication number: WO2018042459A1
Application number: PCT/IN2017/050374
Authority: WO
Inventors: Murugavel MUTHUKUMARASAMY
Original assignee: Muthukumarasamy Murugavel
Priority date: 2016-09-02
Filing date: 2017-09-01
Publication date: 2018-03-08

Abstract

An adaptive multi-layer traffic steering method that intelligently leverages one or multiple network paths to optimize network communications, based on specified QoS needs as well as link / path quality is proposed. The method may use either a single path forwarding or multi-path transport proxy aggregation mode for each flow (TCP/UDP). The method may involve, for each flow / transport session, dynamically selecting one or more network paths that match the traffic class in terms of needs defined by a policy and that of current network conditions at multiple nodes in the network. The performance parameters considered for optimization could be packet loss, RTT, BW, congestion, relative received signal strength (RSSI) and so on. Advantages of the system and method are that adaptive steering and treatment of traffic is seamless for TCP or UDP based applications that reduces cost and significantly improves the service experience for customers and users.

Description

ADAPTIVE AND SEAMLESS TRAFFIC STEERING AMONG MULTIPLE

PATHS BASED ON APPLICATION OOE NEEDS

CROSS-REFERENCES TO RELATED APPLICATIONS

[0001] This application is a complete specification of and claims priority to Indian provisional patent application no. 201641030165 dated September 02, 2016.

FIELD OF THE INVENTION

[0002] The disclosure relates generally to switching in data networks and in particular to a method of adaptive multi-layer traffic steering method that intelligently leverages one or multiple network paths to optimize network communications.

DESCRIPTION OF THE RELATED ART

[0003] Routers and gateways (CPE) either used by enterprises or for residential use today completely rely on routing protocols such as OSPF, BGP etc. or other simple network probing tools to detect link failures and re-route the traffic from one path to another path. The above approach has many disadvantages. Routing protocols and probing tools widely used today are only capable of detecting hard link up/down status and cannot detect service degradation in terms of SLA KPIs (packet loss, RTT, throughput etc.) which are key quality parameters in WAN communications. Even for hard interface up/down events, convergence time for routing protocols range from at least 20 msec to a few minutes, with detrimental effect on performance and user experience. Routing protocols and policy based routing in many scenarios cannot work together and even in simpler scenarios, the configuration could be too complex to deploy and troubleshoot.

[0004] Currently available routing protocols are thus not suitable for branch / home gateway in today's highly digitized, connected and cloud world, where connectivity and user experience are of critical importance. Hence, in recent times there Software Defined Networks (SDN) and SD-WAN technologies which incorporates the capabilities of basic monitoring of quality of paths and steering of traffic.

[0005] Various methods have been disclosed that may perform traffic steering in a communication network. In the published application US20150016249A1 an edge- gateway multipath method is disclosed that communicatively couples an edge device with a cloud-computing service in a cloud-computing network. The US patent US8452846B2 introduces a predicable quality shared distributed memory process. Another US patent US8509089B2 discloses a path selection method for forwarding data packets in a wireless communication system. An ant-based method for discovering a network path that satisfies a quality of service (QoS) requirement is disclosed in the patent US7466655B 1. A routing algorithm to define the instantaneous best path to the destination is discussed in US7561526B2. The invention discloses novel systems, devices and methods to steer traffic between a sender device and one or more receiver devices, that may overcome many of the disadvantages discussed above.

SUMMARY OF THE INVENTION

[0006] The invention in its various embodiments discloses a method of optimizing communication between a sender host and at least one receiver host in a network that may include packets of data as network traffic. The optimized communication envisages providing a user better QOE through minimizing various kinds of issues characterized by network latency, packet loss, and other KPIs as further set forth herein. The sender may be connected to the receiver host via a number of possible network paths. In various embodiments each network path is associated with a quality of service (QoS) configured for each application. In various embodiments the method may include identifying predefined policy requirements. A multi-path routing algorithm may query and determine the network quality of service (QoS) correspond to each of the number of possible network paths. The method may then identify a network path or a combination of network paths that may provide the required QoS parameters to meet the predefined policy requirements. In various embodiments the network traffic is routed through the identified network path or combination of paths and the traffic is delivered to the at least one receiver network host.

[0007] In some embodiments identifying predefined policy requirements include receiving the network policy on traffic steering rules from the management plane. The multi-path routing algorithm may initialize and configure an adaptive Performance Monitoring sub-system (APMS) and a Path Quality DB (PQDB) sub-systems based on the received network policy. Further appropriate forwarding policies and rules are configured in the data-plane.

[0008] In various embodiments querying and determining the QoS includes querying the PQDB for link / path quality information and evaluates the defined policy rules against each network. Further new forwarding rules are installed in the data plane to choose an egress interface. [0009] In various embodiments, to identify a network path includes classifying and marking the packet of data based on the defined network policy. Based on packet marking a policy forwarding table may be selected. Further, the right egress interface / path may be identified and packet may be routed. [0010] In some embodiments to identify a combination of network paths includes evaluating and identifying another egress /path link that may fulfill QoS requirements. Previously installed forwarding rule may be removed and a new forwarding rule may be installed to forward the packet to a multi-path transport proxy client/server and the multi- path transport proxy client/server splits the single TCP/UDP flow into multiple sub- flows.

[0011] In various embodiments the performance parameters are one or more of packet loss, round-trip time (RTT), bandwidth (BW), network congestion relative received signal strength (RSSI) or minimum BW availability. In various embodiments a multi- path protocol is used to select a combination of two or more network paths, that together provide the required QoS parameters.

[0012] The method in some embodiments may include selecting a combination of two or more network paths that may include identifying a multi-path intermediate transport proxies in the network path between the sender host device and the receiver host device and adaptively changing the multi-path intermediate transport proxies based on link or path quality across a single TCP or UDP flow. In some embodiments routing the network traffic through a combination of network paths comprises aggregating the traffic before delivering to the receiver host.

[0013] In various embodiments a smart multi-path traffic steering agent(SMA) device incorporating the method of optimizing communication between a sender host and one or more receiver hosts in a network is disclosed. In various embodiments the SMA device is configured to identify a network path or a combination of network paths that provide the required QoS parameters to meet the predefined policy requirements. [0014] A system for optimizing communication between a sender host and one or more receiver host in a network is disclosed in various embodiments. The system may include at least a sender host and one or more receiver hosts connected in the network, one or more intermediaries connecting the sender host and the one or more receiver hosts in the network, one or more connector devices connecting the sender host or the one or more receiver hosts to the network via optimized communication,

[0015] In various embodiments the one or more connector devices configured to implement the method may identify predefined policy requirements. In some embodiments the system may query and determine the network quality of service (QoS) for an application that correspond to each of the number of possible network paths. The system may then identify a network path or a combination of network paths that may provide the required QoS parameters to meet the predefined policy requirements. Further the network traffic is routed through the identified network path or combination of paths and delivered to the at least one receiver network host.

[0016] In some embodiments, the one or more intermediaries are internet service providers, data center or VPN service providers that may provide MPLS VPN based services. In various embodiments, the one or more connector devices are smart multi- path traffic steering agent (SMA) devices. In various embodiments, the SMA device is part of a router, a host device, a local Edge / Gateway device, an implicit/explicit proxy within the service provider's data center (DC) or customer's DC or a cloud gateway, a private or public cloud server, or a custom-built device. In some embodiments, the SMA is an independent device connected to a host or a router or a gateway.

[0017] In some embodiments, each SMA device in the network is configured to optimize communication. [0018] This and other aspects are set forth herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The invention has other advantages and features which will be more readily apparent from the following detailed description of the disclosure and the appended claims, when taken in conjunction with the accompanying drawings, in which: [0020] FIG. 1 illustrates the system for intelligent multi-path traffic steeringin the

transport session during a TCP/UDP flow.

[0021] FIG. 2 shows the method of intelligent traffic steering by multi-path routing and/or aggregation.

[0022] FIG. 3 illustrates the adaptive multipath forwarding for SMA capable host and gateway.

DETAILED DESCRIPTION

[0023] While the invention has been disclosed with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the invention. In addition, many modifications may be made to adapt to a particular situation or material to the teachings of the invention without departing from its scope.

[0024] Throughout the specification and claims, the following terms take the meanings explicitly associated herein unless the context clearly dictates otherwise. The meaning of "a", "an", and "the" include plural references. The meaning of "in" includes "in" and "on." Referring to the drawings, like numbers indicate like parts throughout the views. Additionally, a reference to the singular includes a reference to the plural unless otherwise stated or inconsistent with the disclosure herein.

[0025] Throughout the specification acronyms WAN, OSPF, SLA, KPI, SP, QOE, QOS, SDN are used. WAN stands for Wide Area Networks, OSPF means Open Short Path First, SLA stands for Service Level Agreement, KPI means Key Performance Indicators, SP means service provider, QOE means quality of experience, QOS means quality of service, and SDN means software defined network.

[0026] The invention in its various embodiments proposes an adaptive multi-layer traffic steering method that intelligently leverages one or multiple network paths to optimize network communications, based on specified QoS needs as well as link / path quality. The optimized communication envisages providing a user better QOE through minimizing various kinds of issues characterized by network latency, packet loss, and other KPIs as further set forth herein. The method may use either a single path forwarding or multi-path transport proxy aggregation mode for each flow (TCP/UDP). [0027] In one embodiment, a system 100 as shown in FIG. 1 for optimizing communication between a sender host 101 and one or more receiver hosts 103 present in a communication network is disclosed. As illustrated in FIG. 1, the system comprises hosts at various endpoints 101, 103 that are connected via a network comprising one or more intermediaries 111, 113, 115. Each host 101, 103 may be connected to the network via connector devices that are configured to implement a smart multi-path traffic steering algorithm that may meet predefined policy requirements. In various embodiments, the connector devices may include a custom built devicel30-l, 130-2 configured with the smart multi-path traffic steering algorithm or devices functionalized with the smart multi-path traffic steering algorithm that may include a host device 170, a router 120, a local Edge / Gateway device 180, an implicit/explicit proxy within the service provider's data center (DC) 190 or customer's DC or a cloud gateway 185. In various embodiments the connector devices 120, 130-1, 130-2, 170, 180, 185, 190 are configured to identify predefined policy requirements of the sender host end. The connector device 120, 130-1, 130-2, 170, 180, 185, 190, may then query and determine the network quality of service (QoS) that correspond to each of the number of possible network paths. Further, the connector device 120,130-1, 130-2, 170, 180, 185, 190 may identify a network path or a combination of network paths that may provide the required QoS parameters to meet the predefined policy requirements configured for the application. The system may then steer traffic to the receiver host 103 through either the identified network path or combination of network paths.

[0028] In various embodiments the one or more intermediaries are internet service providers or VPN service providers that may provide MPLS VPN based services. In various embodiments the one or more connector devices are smart multi-path traffic steering agent (SMA) devices 130-1, 130-2 that are configured to implement a smart multi-path traffic steering algorithm that may optimize communication between a sender host 101 and a receiver host 103. The SMA devices are configurable by policy language to route traffic according to pre-defined QoS parameters required in each path for the particular application. In various embodiments the multi-path traffic steering policies may be defined locally within the SMA or globally by the controller 150 and communicated to the SMA. In some embodiments the network path selection routing decision may also include considering periodic traffic intelligence shared by the controller that may include data gathered by the SDN or network controllers.

[0029] In some embodiments the SMA device may gather QoS parameters by enabling traffic probes 140-1, 140-2 located in the edge device that are integrated with performance monitoring devices. The traffic probes 140-1, 140-2 may in some embodiments be integrated with the SMA. In some embodiments the traffic probes 140- 1, 140-2 may be enabled either in active or passive mode. In one embodiment while enabled in active mode the probes may generate traffic to test bandwidth inside networks to measure latency. In another embodiment while the traffic probes are enabled in passive mode the probes may periodically capture traffic and may upload packets to the server for analysis.

[0030] The SMA in various embodiments may work as an overlay on top of the routing protocol, that may handle the underlying network reachability. The routing protocol may use distance vector, path vector protocols or link state protocol to determine the best path a packet of data could take to reach a receiver host. In various embodiments the smart multi-path traffic steering algorithm may identify an appropriate path for a traffic profile that may override a best path decision taken by the routing protocol

[0031] The SMA device 130-1, 130-2 in various embodiments, may be a part of an edge device, a switch, a router, an implicit or explicit transport proxy or an application proxy. The SMA device in various embodiments may also be a custom-built device configured to implement the smart multi-path routing algorithm for intelligent routing as disclosed herein. In some embodiments, an SMA may also be configured on desktop or mobile devices that may adapt to the reliability requirements of the devices.

[0032] The SMA device 130-1, 130-2 disclosed in FIG. 1 herein can in some embodiments be collocated or be a part within an end point device 170. In some embodiments the SMA could be a part or collocated within a router 160 or gateway 180. In various embodiments the SMA functionalities complements the routing protocol with better network resiliency and service experience.

[0033] In some embodiments the SMA 130-1, 130-2 may be an independent device connected to either a host or router / gateway. In other embodiments the SMA can be deployed as a standalone, custom-built or virtual device as part of service provider network or a private / public cloud.

[0034] The invention in its various embodiments 200 discloses a method that may optimize communication between a sender host and at least one receiver host in a communication network. The sender host and the receiver host may include a smart multi-path traffic steering algorithm and are connected via a number of network paths. The method 200 comprises the steps 201 to 207 as illustrated in FIG. 2. In step 201 network traffic is initiated from a sender host to a receiver host that are part of a communication network. The network traffic may include packets of data to be sent between the sender host and the receiver host. In some embodiments, each network path includes a quality of service (QoS) associated with each application. In step 202 the predefined policy requirements of the networks are identified. In some embodiments, the network policy on traffic steering rules are received from the management plane. An adaptive Performance Monitoring sub-system (APMS) and a Path Quality DB (PQDB) sub-systems based on the received network policy are initialized and configured. Further appropriate forwarding policies and rules are configured in the data-plane. In step 203 a smart multi-path traffic steering algorithm may query and determine the network QoS corresponding to each of the number of possible network paths that may connect the sender host to the receiver host. In various embodiments the algorithm may query the PQDB for link or path quality information and may evaluate the defined policy rules against each network. In some embodiments the path quality may be queried for a set of network nodes between the reference node comprising the multi-path traffic steering algorithm to a pre-configured gateway or the intermediate / destination node. The algorithm in some embodiments may also install new forwarding rules in the data plane to choose an egress interface.

[0035] The smart multi-path traffic steering algorithm in step 204 may try to identify a single network path that may be sufficient to provide the required QoS parameters. In one embodiment if a single network that may meet the predefined policy requirement is identified the smart multi-path traffic steering algorithm in step 205A may route traffic to the receiver host through the single network path. In some embodiments to identify a network path the algorithm may classify and mark the packets of data based on the defined network policy. Based on packet marking a policy forwarding table may be selected and the egress interface or path to route the packet identified.

[0036] In another embodiment if a single network path does not suffice to meet the required QoS parameters, then in step 205B the smart multi-path traffic steering algorithm selects a multi-path protocol that identifies a combination of network paths that provide the required QoS parameters. Step 205B may evaluate and identify another egress /path link that fulfills the QoS requirements. In some embodiments when another path is identified, a previously installed forwarding rule may be removed and a new forwarding rule is installed to steer the packet over the identified egress interface. In some embodiments the packet may be forwarded to a multi-path transport proxy client, that is connected to another SMA agent running a multi-path transport proxy server. The multi-path transport proxy client / server may split the single TCP/UDP flow into multiple sub-flows and steer the traffic based on link characteristics such as congestion. The multi-path transport proxy client / server may in some embodiments are fully configurable / customizable. In various embodiments the multipath transport proxy may include WAN optimization functionality viz. data deduplication, compression etc

[0037] In step 206 the multi-path transport session is initiated towards the receiver host. Step 203 is repeated to identify the next network path that may provide the required QoS parameters. If the single path would sufficiently meet the QoS parameters multi-path transport session could be discontinued and the traffic may be steered onto a single path. Further, in step 207 the traffic routed via the multi-path protocol is aggregated from the multiple paths and passed on to the receiver host. The return traffic is handled in a similar fashion making the communications seamless at the hosts A and B.

[0038] In some embodiments, in step 204, if a single interface / path does not sufficiently provide the required QoS for a single TCP/UDP flow, the method in step 205B may insert a transport proxy functionality before the multi-path selection is made. The transport proxy client may encapsulate the packet and forward to a multi-path capable transport proxy at the far end with which the SMA has already established a link. The SMA and the multi-path transport capable proxy configured with SMA features may then steer the packet / data between one or more paths adaptively as per the link quality and as per policy defined. In various embodiments the method is also configured to deactivate the proxy functionality and may forward the traffic to a multi-path capable proxy. The method may switch to a regular single application flow on multiple paths or single application flow on a single path.

[0039] In various embodiments in step 202 the required QoS parameters may include one or more of packet loss, round-trip time (RTT), bandwidth (BW) or network congestion. In various embodiments QoS key performance indicators (KPI) may include relative received signal strength (RSSI) for Wi-Fi and 3G/4G interfaces and may include minimum BW availability defined per application. In various embodiments the path quality measurements may be either one-way or two-way based on the configuration and end-application.

[0040] In some embodiments the smart multi-path traffic steering algorithm, while selecting a combination of two or more network paths in step 205B may identify multi- path intermediate transport proxies in the network path between the sender host device and the receiver host device. Further, the smart multi-path traffic steering algorithm may adaptively change the multi-path intermediate transport proxies based on link or path quality across a single TCP or UDP flow.

[0041] In various embodiments routing the network traffic in step 207 through a combination of network paths comprises aggregating the traffic before delivering to the receiver host.

[0042] In one embodiment a smart multi-path traffic steering agent(SMA) device may be incorporated with the smart multi-path traffic steering algorithm as detailed with reference to FIG. 2. The SMA device may route traffic between a sender host and one or more receiver hosts that are present in a communication network. The SMA device in various embodiments may be configured to identify a network path or a combination of network paths that meet the predefined policy requirements.

[0043] In various steps of the method, the policy may define the required QoS in terms of various constraints such as business criticality, cost of paths/circuits or other QoS parameters. The method may involve, for each flow / transport session, dynamically selecting one or more network paths that match the traffic class in terms of needs defined by a policy and that of current network conditions at multiple nodes in the network. In some embodiments, each traffic class may include parameters comprising latency, packet loss, jitter and minimum bandwidth as QoS metrics. A real-time VoIP class may have 200 msec latency, 0.5% packet loss, 50 msec jitter and 50 kbps minimum bandwidth as profile parameters, for example. [0044] In various embodiments, the method as disclosed in FIG. 2 could be repeated at intermediate levels and traffic forwarding decisions may be taken at each of the SMA devices In some embodiments, for a transport session lasting for a specific duration of time, the SMA devices may continually evaluate link / path quality and apply the dynamic steering at any point of time in the course of the session or flow. In some embodiments the multi-path aggregation method may be performed in a hierarchic manner, i.e. the aggregation is done for part of the traffic routed via multi-path protocol along a network path that is aggregated later with other packets belonging to the same network data stream. The method is repeatedly implemented to route traffic in any direction in general, in a network.

[0045] Advantages of the system and method are that adaptive steering and treatment of traffic is seamless for TCP and UDP based applications that significantly improves the service experience for customers and users. Another advantage of the method is that traffic forwarding decisions may be taken by the application running in the CPE/Edge devices. Also, the system overcome the limitations of last-mile connectivity issue to support the application requirements.

[0046] The advantages of the disclosed concept include a) lower decision making delays and hence reduced latency (RTT) for the traffic, b) uninterrupted connectivity and higher QoS - even if a cloud controller associated with one service provider goes down (service unavailable), the CPE/Edge continues to operate optimally. The system and method also ensure c) lower resource requirements on the cloud gateway and hence the method is highly cost-effective.

[0047] Examples

[0048] Example 1: SMA Network Configurations [0049] In a possible SMA device configurations in a network as shown in FIG. 1 Hosts 101 and Host 103 were connected to SMA gateway devices at the respective end. Another example is shown in FIG. 3 where Host A and Host C themselves were configured as SMAs and the method was implemented. Host B without SMA was connected to an SMA device incorporated in the gateway, that is configured to implement the adaptive multipath aggregation. Thus, traffic between any of the hosts A, B or C could be routed using the disclosed methods of the invention.

[0050] Example 2: Policy Language

[0051] Traffic Type A (Business Critical) was steered across Internet / MPLS - SP X link while traffic type B was steered towards the Internet - SP Y link.

Given Type A is business Critical, the below policy configuration was used:

i. When either Internet / MPLS link from SP X went down or service was degraded, Smart Multi path routing agent (SMA) steered the traffic towards Internet link from SP Y.

ii. Additionally, SMA @ GW was configured in such a way that Type A traffic was steered on both links from SP X and SP Y simultaneously by switching from plain IP routing /forwarding to Multi-path transport mode where traffic aggregation happened at the other end.

iii. Optionally, when the individual links/paths were sufficient to meet the requirements, SMA steered the traffic from multi-path transport proxy mode to regular IP routing mode.

iv. Also, type B traffic was steered only to SP Y link and not to the link from SP X. If Type A used link from SP Y, type B was given low priority.

[0052] Example 3: Smart multi-path traffic steering algorithm for adaptive Traffic Steering

[0053] The adaptive traffic steerer (ATS) includes the following components: [0054] 1. Adaptive Performance Monitoring sub-system (APMS)

[0055] APMS may be responsible for adaptively measuring the link & path quality of various types of interfaces based on implicit and explicit configurations. The sub-system includes intelligent active and passive probes to measure link/path quality in simple systems to suitably adapted standards based implementations such as Y.1731 operation, administration and maintenance (OAM) or Two-Way Active Measurement Protocol (TWAMP) specifically active measurements. The sub-system may continuously populate the Path Quality DB (PQDB) for configured intervals say, every 500 ms.

[0056] 2. Path Quality DB (PQDB) [0057] Information in an embedded real-time DB populated by APMS, is maintained across multiple dimensions. It may provide traffic intelligence and recent performance of paths / links to other sub-systems. The data is gathered on various parameters including link/interface, specific path, application type or a domain name.

[0058] 3. Adaptive Traffic Steerer (ATS) The Control plane sub-system at time, t₀ may be responsible for: i) . Receiving the Business / Network policy on traffic steering rules from Management plane / Orchestrator. ii) . Initializing & configuring APMS and PQDB sub-systems based on the received Business / Network policy. iii). Configuring appropriate forwarding policies and rules in the data-plane.

[0059] Pre-Traffic Condition:

[0060] ATS may periodically query the PQDB for link / path quality information and evaluates against each network policy rules defined. If a particular application traffic class has a pre-defined requirements of say. 200 msec RTT and 0.5 packet loss and default preferred path doesn't meet the requirement as per the previous 3 to 5 samples, it appropriately installs new forwarding rules in data plane in such a way, a new egress interface is chosen. [0061] Scenario 1 : Packet Arrives and default preferred path selected

1. The packet enters ingress interface. The packet is classified and marked based on the network policy defined.

2. Based on the packet marking, appropriate policy forwarding table may be selected.

3. Further, the right egress interface / path may be identified and packet may be routed.

[0062] Scenario 2: Preferred Link / Path for a Traffic breaches QoS requirements

1. ATS may evaluate and decide that another egress /path link fulfills QoS requirements. 2. Previously installed forwarding rule may be removed and a new forwarding rule installed so that packets of flow are steered over a newly identified egress interface.

[0063] Scenario 3: No single path can fulfill QoS requirements

1. Dynamically insert forwarding rule such that packet is punted to a multi-path transport proxy client, that is already connected to another SMA agent running a multi-path transport proxy server.

2. The multi-path transport proxy client / server themselves may split a single TCP/UDP flow into multiple sub-flows (per-packet load balancing) and steer the traffic based on link characteristics such as congestion and they are fully configurable / customizable. [0064] During the course of TCP/UDP flow and periodic ATS evaluation of PQDB against network policy/rules, if a single link / path seem sufficient to fulfill the needs of application QoS rules, any previously installed forwarding rules are deleted so that regular Layer 3 forwarding happens.

Claims

WE CLAIM:

1. A method of optimizing communication between a sender host and at least one receiver host in a network, wherein the network traffic comprises packets of data, the sender connected to the receiver host via a number of possible network paths, each network path associated with a quality of service (QoS), comprising:

identifying predefined policy requirements;

querying and determining the network quality of service (QoS) corresponding to each of the number of possible network paths;

identifying a network path or a combination of network paths that provide the required QoS parameters to meet the predefined policy requirements;

routing the network traffic through the identified network path or combination of paths; and

delivering the traffic to the at least one receiver network host.

2. The method of claim 1 wherein identifying predefined policy requirements comprises

receiving the network policy on traffic steering rules from Management plane;

initializing and configuring an adaptive Performance Monitoring subsystem (APMS) and a Path Quality DB (PQDB) sub-systems based on the received network policy;

configuring appropriate forwarding policies and rules in the data-plane.

3. The method of claim 1 wherein querying and determining the QoS comprises querying the PQDB for link / path quality information;

evaluating the defined policy rules against each network; installing new forwarding rules in data plane to choose an egress interface.

4. The method of claim 1 wherein identifying a network path comprises

classifying and marking the packet of data based on the defined network policy;

selecting a policy forwarding table based on packet marking; identifying the egress interface or path to route the packet.

5. The method of claim 1 wherein identifying a combination of network paths comprises

evaluating and identifying another egress /path link to fulfill QoS requirements;

removing previously installed forwarding rule;

installing a new forwarding rule to forward the packet to a multi-path transport proxy client/server and wherein the multi-path transport proxy client/server splits the single TCP/UDP flow into multiple sub-flows.

6. The method of claim 1 wherein the performance parameters are one or more of packet loss, round-trip time (RTT), bandwidth (BW), network congestion, relative received signal strength (RSSI) or minimum BW availability.

7. The method of claim 1 wherein a multi-path protocol is used to select a combination of two or more network paths, that together provide the required QoS parameters.

8. The method of claim 1 wherein routing the network traffic through a combination of network paths comprises aggregating the traffic before delivering to the receiver host.

9. A smart multi-path traffic steering agent(SMA) device incorporating the method of claim 1 to route traffic between a sender host and one or more receiver hosts in a network, wherein the SMA device is configured to identify a network path or a combination of network paths that provide the required QoS parameters to meet the predefined policy requirements.

10. A system for optimizing communication between a sender host and one or more receiver host in a network comprising

at least a sender host and one or more receiver hosts connected in the network;

one or more intermediaries connecting the sender host and the one or more receiver hosts in the network;

one or more connector devices connecting the sender host or the one or more receiver hosts to the network via optimized communication, the one or more connector devices configured to implement a method configured to

identify predefined policy requirements;

query and determine the network quality of service (QoS) that correspond to each of the number of possible network paths;

identify a network path or a combination of network paths that provide the required QoS parameters to meet the predefined policy requirements;

route the network traffic through the identified network path or combination of paths; and deliver the traffic to the at least one receiver network host.

11. The system of claim 10 wherein the one or more intermediaries are internet service providers, data center or VPN service providers that may provide MPLS VPN based services.

12. The system of claim 10 wherein the one or more connector devices are smart multi-path traffic steering agent(SMA) devices.

13. The system of claim 12, wherein each SMA device in the network is configured to optimize communication.

14. The system of claim 12 wherein the SMA is part of a router, a host device, a local Edge / Gateway device, an implicit/explicit proxy within the service provider's data center (DC) or customer's DC or a cloud gateway, a private or public cloud server, or a custom-built device.

15. The system of claim 12 wherein the SMA is an independent device connected to a host or a router or a gateway.