What is the difference between PermGen and Metaspace?

Bye, Bye PermGen, Hello Metaspace

PermGen has been completely removed.

Metaspace garbage collection - Garbage collection of the dead classes and classloaders is triggered once the class metadata usage reaches the MaxMetaspaceSize.

The space Metadata was held is no longer contiguous to the Java heap, The metadata has now moved to native memory to an area known as the Metaspace.

In Simple words,

Since the class metadata is allocated out of native memory, the max available space is the total available system memory. Thus, you will no longer encounter OOM errors and could end up spilling into the swap space.

The removal of PermGen doesn’t mean that your class loader leak issues are gone. So, yes, you will still have to monitor your consumption and plan accordingly, since a leak would end up consuming your entire native memory.

Some other articles, with analysis: Link1, Link2 , and this


In short, Metaspace size auto increases in native memory as required to load class metadata if not restricted with -XX:MaxMetaspaceSize


The main difference from a user perspective - which I think the previous answer does not stress enough - is that Metaspace by default auto increases its size (up to what the underlying OS provides), while PermGen always has a fixed maximum size. You can set a fixed maximum for Metaspace with JVM parameters, but you cannot make PermGen auto-increase.

To a large degree it is just a change of name. Back when PermGen was introduced, there was no Java EE or dynamic class(un)loading, so once a class was loaded it was stuck in memory until the JVM shut down - thus Permanent Generation. Nowadays classes may be loaded and unloaded during the lifespan of the JVM, so Metaspace makes more sense for the area where the metadata is kept.

Both of them contain the java.lang.Class instances and both of them suffer from ClassLoader leaks. Only difference is that with Metaspace default settings, it takes longer until you notice the symptoms (since it auto increases as much as it can), i.e. you just push the problem further away without solving it. OTOH I imagine the effect of running out of OS memory can be more severe than just running out of JVM PermGen, so I'm not sure it is much of an improvement.

Whether you're using a JVM with PermGen or with Metaspace, if you are doing dynamic class unloading, you should to take measures against classloader leaks, for example by using my ClassLoader Leak Prevention library.