Shulou(Shulou.com)06/02 Report--

This article introduces the relevant knowledge of "how to understand the Heap of the Java virtual machine". Many people will encounter such a dilemma in the operation of actual cases, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

In the Java virtual machine, I am a powerful butler, they are very afraid of me, especially those Java objects, I put them in a "prison" called Heap, strict management, life and death in control.

The Chinese translate Stack into "stack", Heap into "stack", and some people translate Stack into "stack". Alas, I don't know what they think, but after all these years, it's good for you to understand.

I happen to garbage collect Java objects in Heap, and this "heap" always reminds me of garbage dumps.

When it comes to garbage collection, this is really a big burden, the reason is very simple: those people who write Java programs just give the objects to new and throw them into Heap, but never delete them, and the responsibility for deleting these objects falls on me. How can I not manage them strictly?

Sometimes I envy C and Clippers because they have to allocate and release memory manually, and programmers take the blame when things go wrong.

In my case, if these objects are allowed to act recklessly, my low-capacity Heap "prison" that cannot be changed after the Java virtual machine is started will soon be filled, so I have to send my right-hand men to find and clean up those unused Java objects and release the space they occupy.

In order to find these troublemakers, I invented an algorithm called "reachability analysis". This algorithm is probably already known to most people, so I won't talk about it any more. The following picture illustrates the idea behind it. Smart you can see at a glance that orange objects are unreachable objects and can be recycled.

1. Heap Prison

All right, now let's talk about the Heap "prison" I manage in detail.

You can think of it as a large space, and for ease of management, I divided the Heap "prison" into multiple areas and moved those Java objects around.

My rule is: the new guys are going to stay in the new generation, and the new generation can't live any longer, so I send cleaners to Minor GC. If they can't live after recycling, then drive the old guys to a nursing home (old age).

I will set an age counter for every Java object in Heap. Every time the Java object survives GC, it will increase its age by 1. If it is too old, sorry, please enter the nursing home (old age). In fact, I will also make a dynamic age judgment, press the table here.

You may wonder why there are strange areas like Eden, Survivor1 and Survivor2 in the Cenozoic era.

That's because I want to implement a so-called "replication" algorithm here.

In the earliest days, I divided an area of memory into two areas of the same size, using only one of them at a time.

When area 1 is used up, I will do garbage collection and move what is still alive to another area.

Note: after moving in, they all live close to each other, so that the red fragments that have been cleaned up will be flattened into a large space for subsequent use, especially for large objects.

Using the two areas upside down in this way, although efficient and free of debris, the wasted space is huge: only half of it can be used at a time.

Later, humans found that most of the objects in the new generation could not live for long, and basically they were almost deleted in a single garbage collection.

So we improved this half-copy algorithm to divide the new generation into three parts: Eden, Survivor1, and Survivor2, with a ratio of 8:1:1.

Use only Eden and one of the Survivor at a time. When garbage collection, copy the living objects in these two areas to another Survivor. If the Survivor doesn't fit, please go to a nursing home (old age).

If unfortunately, even the nursing home is full, then have to do a Full GC, this is a very slow operation, you'd better pray that it does not happen frequently.

2. Apart from the "prison", there is a bright future.

Although I can dominate in Heap prison, sometimes I have to get in touch with the world outside the prison.

Once I wanted to send data through Socket. I obviously prepared the data, which was in my Heap, but the boss of JVM actually copied a copy of the data to the memory outside the Heap before it could be sent through Socket.

I asked him what this was all about and why he had to do it. Was he worried about me, the butler of Heap prison?

The boss of JVM said that he is really worried that the Socket at the bottom of others is written in C language, focusing on the address of physical memory. When you garbage collect Java objects, you move them around between Eden, Survivor, and old times, and the address of the object will also change. How can I tell people which address data to send?

Think of the same reason, there are gains and losses, you programmers do not have to manage memory, but the bottom has to deal with memory, and there is an extra process: Copy.

The boss also said: "maybe you don't know, in addition to your Heap prison, I actually have a place called 'Off-Heap memory' in the Java process, and the data will be copied here." In order to distinguish it from you, I call it out-of-stack memory. "

I didn't expect that there was an enclave that I couldn't control!

But it doesn't compete with me, so let it go.

But a few days later, boss JVM brought me a "surprise" again.

"it's too troublesome to copy data," he said. "I figured out a way to allocate a piece of memory that belongs to Off-Heap directly in the Java code."

I think the scalp is blurred: "allocate memory directly outside the heap? how on earth do you allocate it?"

The boss gave me a piece of code: "look, this is not the allocation of 128m of heap memory? read and write to this buffer will be directly written to the out-of-heap memory, no longer need to be copied by you."

ByteBuffer buffer = ByteBuffer.allocateDirect (1024024128)

Skycto JEEditor: generate function code with one click and program directly to the interface.

Damn interface-oriented programming, the out-of-heap memory allocated by this ByteBuffer is used like a normal Java object, with no idea whether it is in or out of the heap.

It's over. I can't control this piece of memory at all.

The boss saw that I was not in the right mood and comforted: "this buffer is also a Java object. It is stored in your Heap, but it stores the information of the 128m memory."

That's more like it! Since it is a Java object, it has to be put in my Heap prison and controlled by me!

As you can imagine, when this object is garbage collected, the direct memory it points to will be released.

I suddenly had an evil idea: if there are more and more such objects and have not been garbage collected, then the corresponding direct memory cannot be released, and then Out of Memory?

The boss seems to see through my mind: "for these objects, you have to be very careful and make sure they can be released."

The function of directly allocating out-of-heap memory has been officially launched, and I found that allocating out-of-heap memory is a little slower than in-heap memory. But what I didn't expect is that it is especially suitable for those scenarios where the allocation times are small and the read and write operations are very frequent. So it is warmly welcomed by these communication systems such as Netty.

To reduce the overhead of creating out-of-heap memory, Netty also introduces object pooling, just like database connection pooling, allocating some out-of-heap memory and then constantly reusing them.

I didn't expect that out-of-heap memory could play so many tricks, but I was relieved to think that they were still Java programs and had to be wrapped in Java objects.

This is the end of the content of "how to understand the Heap of the Java virtual machine". Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.