Test Streaming Analytic Function Using Local Test Server
This example shows how to use the development version of MATLAB® Production Server™ to test a streaming analytic function before deployment to MATLAB Production Server. MATLAB Compiler SDK™ includes the development version of MATLAB Production Server, which you can use as a local test server for testing and debugging application code before deploying it to enterprise systems.
You must have Streaming Data Framework for MATLAB Production Server installed on your system. For more information, see Install Streaming Data Framework for MATLAB Production Server.
You must have a running Kafka® server where you have the necessary permissions to create topics. The example assumes that the network address of your Kafka host is
Write Streaming Analytic MATLAB Function
For testing purposes, use the sample MATLAB functions
located in the
is the root
folder of support packages on your system. To get the path to this folder, use this
Later, you test the
recamanSum deployable archive using the local
Create Sample Streaming Data
Prepare for testing by creating sample data and writing the data to a Kafka stream. For this example, you create a 1000-element Recamán sequence and write
it to a Kafka topic
To create the Recamán sequence, you can use the following
recamanTimeTable function, which is also located in the
recamanTimeTable creates a
timetable containing the first
N elements of a Recamán sequence.
function tt = recamanTimeTable(N) rs = zeros(1,N); for k=2:N n = k-1; subtract = rs(k-1) - n; if subtract > 0 && any(rs == subtract) == false rs(k) = subtract; else rs(k) = rs(k-1) + n; end end incr = seconds(1:N); thisVeryInstant = ... convertTo(datetime, "epochtime", "Epoch", "1970-1-1"); thisVeryInstant = datetime(thisVeryInstant, "ConvertFrom",... "epochtime", "Epoch", "1970-1-1"); thisVeryInstant.TimeZone = "UTC"; timestamp = (thisVeryInstant - seconds(N)) + incr'; key = (0:N-1)'; key = string(key); R = rs'; tt = timetable(timestamp,R,key); end
You can use the following code to create a 1000-element Recamán sequence using the
recamanTimeTable function and write it to the
Kafka topic. The example assumes that you have Kafka host running at the network address
you have the necessary permissions to create topics in the Kafka cluster.
kafkaHost = "kafka.host.com"; kafkaPort = 9092; tt0 = recamanTimeTable(1000); dataKS = kafkaStream(kafkaHost, kafkaPort, "recamanSum_data", Rows=100); try deleteTopic(dataKS); catch, end writetimetable(dataKS, tt0); tt1 = readtimetable(dataKS); if isequal(tt0(1:height(tt1),:), tt1) fprintf(1,"Success writing data to topic %s.\n", dataKS.Name); end stop(dataKS);
Simulate Production Using Local Test Server
To simulate streaming data processing in a production environment, you can run the
recamanSum deployable archive using the development version of
Production Server and process data from the
KafkaStreamobject connected to the
ks = kafkaStream("kafka.host.com",9092,"recamanSum_data");
KafkaStreamobject to write the results of the
recamanSumanalytic function to a different topic called
outKS = kafkaStream("kafka.host.com",9092,"recamanSum_results");
EventStreamProcessorobject that runs the
recamanSumfunction and initializes persistent state with the
Providing a persistent data storage connection name as an input argument is optional. If you do not provide one,
EventStreamProcessorcreates a connection with a unique name to cache the data state between iterations.
esp = eventStreamProcessor(ks,@recamanSum,@initRecamanSum,OutputStream=outKS);
Using the MATLAB editor, you can set breakpoints in the
recamanSumfunction to examine the incoming streaming data when you start the server.
Start the test server.
Doing so opens the Production Server Compiler app with values for the streaming function
recamanSum, the entry point function
streamfcn, and the deployable archive
Start the test server from the app by clicking Test Client, and then Start.
Navigate back to the MATLAB command prompt to start processing events.
In the Production Server Compiler app, you can see that the test server receives data.
From the MATLAB editor, if you had set breakpoints, you can use the debugger to examine the data, state, and results of the function processing. Click Continue to continue debugging or Stop when you have finished debugging.
From the MATLAB command prompt, stop event processing and shut down the server.
Read the processed results from the output stream.
results = readtimetable(outKS);
results = 50×2 timetable timestamp key sum ____________________ ____ ____ 20-Jan-1970 04:55:08 "0" 0 20-Jan-1970 04:55:08 "1" 1 20-Jan-1970 04:55:08 "2" 4 20-Jan-1970 04:55:08 "3" 10 20-Jan-1970 04:55:08 "4" 12 : : : 20-Jan-1970 04:55:08 "45" 1697 20-Jan-1970 04:55:08 "46" 1732 20-Jan-1970 04:55:08 "47" 1814 20-Jan-1970 04:55:08 "48" 1848 20-Jan-1970 04:55:08 "49" 1931 Display all 50 rows.