Network communications with optimized quality

1. A method for configuring a selected data path between a source and a destination in a communications network comprising a plurality of nodes and links between pairs of said nodes, the selected data path comprising selected nodes of said plurality of nodes and selected links between pairs of said selected nodes, the method comprising:

for each link, continuously maintaining a predictive link performance model for predicting a network quality of the link, including:

for each link, repeatedly measuring a plurality of link parameters of the link generating link parameter data;

for each link, repeatedly processing said link parameter data of the link to generate derived link parameter data and storing said derived link parameter data in a memory;

for each link, repeatedly determining a correlation between the link parameters of the link with use of the derived link parameter data of the link;

for each link, repeatedly determining which of the link parameters of the link are the most predictive link parameters of the link with use of the derived link parameter data of the link; and

for each link, repeatedly determining a weighting for each link parameter of the link with use of the derived link parameter data of the link;

receiving, by a gateway server, a network request from the source, the network request including multiple network parameters defining a requested quality level for the selected path;

selecting the selected nodes and the selected links from the plurality of nodes and plurality of links utilizing the predictive link performance model for each of said selected links and values of said multiple network parameters to determine which of the plurality of links satisfy the requested quality level; and

configuring said the selected nodes between said source and said destination such that data packets sent by said source follow the selected path to said destination.

2. The method of claim 1 wherein the predictive link performance models for each of said links are determined using a machine learning algorithm.

3. The method of claim 2 wherein the machine learning algorithm utilizes a weighted combination of the plurality of measured link parameter data.

4. The method of claim 3 where a higher weighting is given to a more predictive link parameter of the plurality of measured link parameters.

5. The method of claim 1 wherein the requested quality level is defined through an API accessed by a computer program executed on the source.

6. The method of claim 1 further comprising monitoring the link quality levels of the plurality of selected links and the path quality level during data transmissions between the source and the destination.

7. The method of claim 1 wherein data received from the source is encapsulated by a gateway server and de-encapsulated by an application server.

8. A system for routing data between a source and a destination over a selected data path in a communications network comprising a plurality of nodes and links between pairs of said nodes, the selected data path comprising selected nodes of said plurality of nodes and selected links between pairs of said selected nodes, the system comprising:

a plurality of servers, a gateway server and an application server, coupled together through the communications network;

said gateway server configured for:

for each link, continuously maintaining a predictive link performance model for predicting a network quality of the link, including:

for each link, repeatedly measuring a plurality of link parameters of the link generating link parameter data;

for each link, repeatedly processing said link parameter data of the link to generate derived link parameter data and storing said derived link parameter data in a memory;

for each link, repeatedly determining a correlation between the link parameters of the link with use of the derived link parameter data of the link;

for each link, repeatedly determining which of the link parameters of the link are the most predictive link parameters of the link with use of the derived link parameter data of the link; and

for each link, repeatedly determining a weighting for each link parameter of the link with use of the derived link parameter data of the link;

receiving, a network request from the source, the network request including multiple network parameters defining a requested quality level for the selected path;

selecting the selected nodes and the selected links from said plurality of nodes and plurality of links utilizing the predictive link performance model and values of said multiple network parameters to determine which of the plurality of links satisfy the requested quality level; and

configuring the selected nodes between said source and said destination such that data packets sent by said source follow the selected path to said destination.

9. The system of claim 8 wherein the predictive link performance models for each of said links are determined using a machine learning algorithm.

10. The system of claim 9 wherein the machine learning algorithm utilizes a weighted combination of the plurality of measured link parameter data.

11. The system of claim 10 where a higher weighting is given to a more predictive link parameter of the plurality of measured link parameters.

12. The system of claim 8 wherein said derived link parameter data is processed by calculating one count for one of a plurality of bins.

13. The system of claim 8 wherein the requested quality level is defined through an API accessed by a computer program executed on the source.

14. The system of claim 8 wherein data received from the source is encapsulated by a gateway server and de-encapsulated by an application server.