What is Generational GC?

Direct Answer

A generational collector groups objects by age because many workloads create objects that become unreachable quickly. It usually collects young regions more often and promotes survivors to older regions. Region names, promotion rules, event names, and pause behavior depend on the collector, so young collections are not guaranteed to be fast or old collections rare.

Code Example

@Service
public class ReportService {
    private final ReportRepository repository; // retained with the service

    public Report generateReport(long userId) {
        var records = repository.findByUserId(userId);
        var builder = new StringBuilder();
        for (var record : records) {
            builder.append(record).append('\n');
        }
        return new Report(builder.toString());
    }
}

The service and repository are usually long-lived, while builder is useful only during this call. That lifetime pattern illustrates why generational collection can help, but source code does not prove where any object is physically placed. Escape analysis, large-object rules, collector choice, and promotion policy can change the result. Use collector logs to observe the actual allocation and promotion behavior.

Common Mistakes & Pitfalls

• Treating minor and major as JVM-wide terms: Event names and collection scope differ by collector; read the collector name and log cause.
• Assuming all allocation starts in Eden: Some collectors are region-based, and large or special objects can follow different paths.
• Using a fixed heap ratio: Adaptive sizing and collector layouts make universal one-third/two-thirds guidance unreliable. Measure allocation, survival, promotion, and the live set.
• Calling generations physical OS regions: They are collector-managed logical areas and are not a portable promise of one contiguous address layout.