Generation of test traffic configuration based on real-world traffic

Abstract

Some embodiments provide a method for generating a test traffic configuration for testing a first network. From a second network, the method receives a set of data streams representing data traffic observed in the second network. The method uses a machine learning engine to analyze the set of data streams in order to determine traffic patterns in the second network. The method generates the test traffic configuration for testing the first network by replicating the traffic patterns of the second network in the first network.

1 - 20 . (canceled) 
     
     
         21 . A method for generating a test traffic configuration for testing a first application in a first network prior to deploying the first application, the method comprising:
 from a first gateway device operating in a second network, receiving a set of data streams representing data traffic that is (i) sent to a second application deployed in the second network and (ii) observed at the first gateway device entering the second network from an external network;   using a machine learning engine to analyze the set of data streams in order to determine traffic patterns for the data traffic entering the second network from the external network; and   based on the determined traffic patterns, generating a test traffic configuration for testing a system comprising a second gateway device and the first application in the first network prior to deployment of the first application.   
     
     
         22 . The method of  claim 21 , wherein:
 the second network is a production network and the data traffic represented by the set of data streams is real-world data traffic received by the first gateway device from client devices accessing the second application via a public network; and   the first network is a test network and the first application is a new release of the second application.   
     
     
         23 . The method of  claim 21 , wherein:
 the second network is a production network and the data traffic represented by the set of data streams is real-world data traffic received by the first gateway device from client devices accessing the second application via a public network; and   the first network is a test network and the first application is another instance of the second application.   
     
     
         24 . The method of  claim 21  further comprising providing the generated test traffic configuration to a testing controller that uses the test traffic configuration to configure a set of test traffic sources. 
     
     
         25 . The method of  claim 24 , wherein the test traffic sources are network endpoints deployed in the first network that generate and send data traffic to the second gateway device based on the generated test traffic configuration. 
     
     
         26 . The method of  claim 21 , wherein:
 the set of data streams comprises (i) a first data stream having a first format and (ii) a second data stream having a second format; and   the method further comprises normalizing the first and second data streams into a same standardized format for consumption by the machine learning engine.   
     
     
         27 . The method of  claim 21 , wherein:
 each of the data streams represents the observed data traffic using a respective plurality of dimensions; and   using the machine learning engine to analyze the set of data streams comprises:
 determining an optimal number of dimensions for the data streams; and 
 reducing dimensionality of the data streams to the determined optimal number of dimensions for further analysis by the machine learning engine. 
   
     
     
         28 . The method of  claim 21 , wherein using the machine learning engine to analyze the set of data streams comprises determining, based on traffic statistics during a time interval during which the first gateway device observed the represented data traffic, characteristics of additional data traffic processed by the first gateway but not represented in the received set of data streams. 
     
     
         29 . The method of  claim 28 , wherein the determined traffic patterns are based on the represented data traffic and the additional data traffic determined by the machine learning engine. 
     
     
         30 . The method of  claim 21 , wherein using the machine learning engine to analyze the set of data streams comprises using a clustering algorithm to determine the traffic patterns for the data traffic entering the second network from the external network. 
     
     
         31 . A non-transitory machine-readable medium storing a program which when executed by at least one processing unit generates a test traffic configuration for testing a first application in a first network prior to deploying the first application, the program comprising sets of instructions for:
 receiving, from a first gateway device operating in a second network, a set of data streams representing data traffic that is (i) sent to a second application deployed in the second network and (ii) observed at the first gateway device entering the second network from an external network;   using a machine learning engine to analyze the set of data streams in order to determine traffic patterns for the data traffic entering the second network from the external network; and   generating, based on the determined traffic patterns, a test traffic configuration for testing a system comprising a second gateway device and the first application in the first network prior to deployment of the first application.   
     
     
         32 . The non-transitory machine-readable medium of  claim 31 , wherein:
 the second network is a production network and the data traffic represented by the set of data streams is real-world data traffic received by the first gateway device from client devices accessing the second application via a public network; and   the first network is a test network and the first application is a new release of the second application.   
     
     
         33 . The non-transitory machine-readable medium of  claim 31 , wherein:
 the second network is a production network and the data traffic represented by the set of data streams is real-world data traffic received by the first gateway device from client devices accessing the second application via a public network; and   the first network is a test network and the first application is another instance of the second application.   
     
     
         34 . The non-transitory machine-readable medium of  claim 31 , wherein the program further comprises a set of instructions for providing the generated test traffic configuration to a testing controller that uses the test traffic configuration to configure a set of test traffic sources. 
     
     
         35 . The non-transitory machine-readable medium of  claim 34 , wherein the test traffic sources are network endpoints deployed in the first network that generate and send data traffic to the second gateway device based on the generated test traffic configuration. 
     
     
         36 . The non-transitory machine-readable medium of  claim 31 , wherein:
 the set of data streams comprises (i) a first data stream having a first format and (ii) a second data stream having a second format; and   the program further comprises a set of instructions for normalizing the first and second data streams into a same standardized format for consumption by the machine learning engine.   
     
     
         37 . The non-transitory machine-readable medium of  claim 31 , wherein:
 each of the data streams represents the observed data traffic using a respective plurality of dimensions; and   the set of instructions for using the machine learning engine to analyze the set of data streams comprises sets of instructions for:
 determining an optimal number of dimensions for the data streams; and 
 reducing dimensionality of the data streams to the determined optimal number of dimensions for further analysis by the machine learning engine. 
   
     
     
         38 . The non-transitory machine-readable medium of  claim 31 , wherein the set of instructions for using the machine learning engine to analyze the set of data streams comprises a set of instructions for determining, based on traffic statistics during a time interval during which the first gateway device observed the represented data traffic, characteristics of additional data traffic processed by the first gateway but not represented in the received set of data streams. 
     
     
         39 . The non-transitory machine-readable medium of  claim 38 , wherein the determined traffic patterns are based on the represented data traffic and the additional data traffic determined by the machine learning engine. 
     
     
         40 . The non-transitory machine-readable medium of  claim 31 , wherein the set of instructions for using the machine learning engine to analyze the set of data streams comprises a set of instructions for using a clustering algorithm to determine the traffic patterns for the data traffic entering the second network from the external network.