Understanding Java Object Memory Layout and the Oop‑Klass Model

The article introduces the Oop‑Klass dual‑model used by HotSpot to represent a Java instance, where an instanceOopDesc object holds the instance data and a Klass object holds type metadata.

It describes the two main components of an object header: the mark word (hash code, lock, age, etc.) and the metadata pointer, which may be a compressed or uncompressed Klass reference depending on the -XX:+UseCompressedClassPointers flag.

The layout of the instance data is explained, including field ordering strategies (0, 1, 2) defined by -XX:FieldsAllocationStyle , and the need for padding to keep the total size a multiple of 8 bytes.

Example Java classes ( People , Person ) are used to illustrate how to compute the total heap size: object header (12 bytes) + instance fields (27 bytes) + padding (1 byte) = 40 bytes, plus the sizes of referenced objects (String and char array), resulting in a total of 96 bytes for a Person instance.

The article shows how to verify the calculation with HotSpot Debugger (HSDB): attach to a running JVM, open the Object Histogram, locate the class, and inspect its Oop structure.

Three programmatic approaches for measuring object size are presented:

Using java.lang.instrument.Instrumentation.getObjectSize() – provides an implementation‑specific approximation.

Using sun.misc.Unsafe.objectFieldOffset() to find the maximum field offset and add the field size, then account for padding.

Using third‑party library com.carrotsearch:java-sizeof (class RamUsageEstimator ) to compute size based on JVM specifications.

Sample code snippets are kept intact:

public class Model {
    public static int a = 1;
    public int b;
    public Model(int b) { this.b = b; }
    public static void main(String[] args) throws InterruptedException {
        Model modela = new Model(2);
        Model modelb = new Model(3);
        Thread.sleep(60 * 1000);
        System.out.println("end");
    }
}

Finally, the article concludes that mastering the JVM object layout is essential for tasks such as memory usage estimation, leak detection, and performance tuning.