Deploy Java applications on Azure Cobalt 200 processors
Introduction
Overview
Create an Azure Cobalt 200 Arm VM
Install Java
Java Baseline Testing
Benchmark using Java Microbenchmark Harness
Next Steps
Deploy Java applications on Azure Cobalt 200 processors
Construct an HTTP response in Java
Before benchmarking with JMH, compile and run a small Java program on your Cobalt 200 VM. The program constructs an HTTP response string in memory and prints it. It does not start a server or perform network I/O.
The single elapsed-time measurement includes JVM startup effects such as class initialization and JIT compilation. Use it to check that the code runs, then use JMH in the next section for warmed-up measurements.
Using a file editor of your choice, create a file named HttpSingleRequestTest.java, and add the content below to it:
public class HttpSingleRequestTest {
public static void main(String[] args) {
long startTime = System.nanoTime();
String response = generateHttpResponse("Tomcat baseline test on Arm64");
long endTime = System.nanoTime();
double durationInMicros = (endTime - startTime) / 1_000.0;
System.out.println("Response Generated:\n" + response);
System.out.printf("Response generation took %.2f microseconds.%n", durationInMicros);
}
private static String generateHttpResponse(String body) {
return "HTTP/1.1 200 OK\r\n" +
"Content-Type: text/plain\r\n" +
"Content-Length: " + body.length() + "\r\n\r\n" +
body;
}
}
Compile and run the Java program
Compile the program and run it with modest heap sizes and the G1 garbage collector:
javac HttpSingleRequestTest.java
java -Xms128m -Xmx256m -XX:+UseG1GC HttpSingleRequestTest
JVM flags explained
- -Xms128m - sets the initial heap size to 128 MB
- -Xmx256m - sets the maximum heap size to 256 MB
- -XX:+UseG1GC - enables the G1 garbage collector designed for low pause times
Sample output
If the program runs successfully, you should see output similar to the following:
Response Generated:
HTTP/1.1 200 OK
Content-Type: text/plain
Content-Length: 29
Tomcat baseline test on Arm64
Response generation took 6780.65 microseconds.
Interpret the result
The response should contain HTTP/1.1 200 OK, a text content type, and the sample body. The content length is valid for this ASCII-only example. For arbitrary text, HTTP content length must reflect the encoded byte count.
The printed time is a single measurement during a fresh JVM run. It varies with class loading, compilation, CPU load, and other activity, and does not measure HTTP server latency or steady-state Java performance. Continue to JMH to measure repeated response construction with warmup.