Server Environment
Linux/RedHat
6 cores
Java 7/8
About application :
We are working on developing a low latency (7-8 ms) high speed trading platform using Java. Multi-leg orders are sent after algo conditions are met
Problem
The orders to the exchange using TCP/IP java.net.Socket APIs (using java.io.OutputStream.write(bytes[] arg0) ). Profiler measurement is records as 5-7 microsec which is very high as per our low latency reqs. We are not made use of setPerformancePreferences() api as suggested in one of the questions posted in stacktrace.
Question
Any alternatives to java.net.Socket to reduce the socket
transmission time?
Any optimization techniques to improve performance
Is setPerformancePreferences() is of any use?
We are not made use of setPerformancePreferences() api
It doesn't do anything and never has. I wouldn't worry about it.
Any alternatives to java.net.Socket to reduce the socket transmission time?
The problem is most certainly not a software one. You can get <8 micro-seconds from Java to Java on different machines, but you need low latency network cards like Solarflare or Mellanox.
If you want fast processing you should consider either a high GHz haswell processor, possibly over clocked to 4.2 or 4.5 GHz or a dual socket Haswell Xeon. The cost of these compared to the cost of trading is not high.
Any optimization techniques to improve performance
Using non-blocking NIO i.e. ByteBuffers and busy waiting on the socket connections. (I wouldn't use Selectors as they add quite a bit of overhead) I would turn off nagle.
For some micro-tuning, use an affinity bound thread on an isolated cpu.
Is setPerformancePreferences() is of any use?
Looking at the source .. I will let you be the judge.
public void setPerformancePreferences(int connectionTime,
int latency,
int bandwidth)
{
/* Not implemented yet */
}
Java 7/8
In term of which version to use, I would start with Java 8 as it has much improved escape analysis which can reduce garbage of short lived objects and thus help reduce latency between GCs and jitter from GCs.
A couple of things come to mind:
JNI: JNI lets you write C code that is ran from your Java code. Critical parts of your Java code that are running to slow can be migrated to C/C++ for improved performance. Work would be needed to first identify what those critical points are and if its worth the effort to move it to C/C++.
Java Unsafe: Wanna get dangerous? Use Java Unsafe to bypass that pesky GC. Here is more info on it. On Github you may find some cool wrapper code to more-safely use Java Unsafe. Here is one. More info.
LMAX Disruptor: Read more about it here. This company is also building a fast trading system in Java. Disruptor allows for faster inter-thread communication.
Bytecode scrutinization: Review your code by looking at the byte code. I have done this for a video game I made and was able to streamline the code. You'll need a good tool for turning your class files into readable bytecode. THis might be the tool i used.
Improved garbage collection: Have you tried using the G1 garbage collector? Or messing around with the older GC's?
Highscalability: This site is full of good info on making code fast. Here is an example that might help.
New API I dont know exactly how to use New API, but it has come up in articles I have read. Here is another article on it. You might need to use it via JNI.
All the operating systems till date have been written in C/C++ while there is none in Java. There are tonnes of Java applications but not an OS. Why?
Because we have operating systems already, mainly. Java isn't designed to run on bare metal, but that's not as big of a hurdle as it might seem at first. As C compilers provide intrinsic functions that compile to specific instructions, a Java compiler (or JIT, the distinction isn't meaningful in this context) could do the same thing. Handling the interaction of GC and the memory manager would be somewhat tricky also. But it could be done. The result is a kernel that's 95% Java and ready to run jars. What's next?
Now it's time to write an operating system. Device drivers, a filesystem, a network stack, all the other components that make it possible to do things with a computer. The Java standard library normally leans heavily on system calls to do the heavy lifting, both because it has to and because running a computer is a pain in the ass. Writing a file, for example, involves the following layers (at least, I'm not an OS guy so I've surely missed stuff):
The filesystem, which has to find space for the file, update its directory structure, handle journaling, and finally decide what disk blocks need to be written and in what order.
The block layer, which has to schedule concurrent writes and reads to maximize throughput while maximizing fairness.
The device driver, which has to keep the device happy and poke it in the right places to make things happen. And of course every device is broken in its own special way, requiring its own driver.
And all this has to work fine and remain performant with a dozen threads accessing the disk, because a disk is essentially an enormous pile of shared mutable state.
At the end, you've got Linux, except it doesn't work as well because it doesn't have near as much effort invested into functionality and performance, and it only runs Java. Possibly you gain performance from having a single address space and no kernel/userspace distinction, but the gain isn't worth the effort involved.
There is one place where a language-specific OS makes sense: VMs. Let the underlying OS handle the hard parts of running a computer, and the tenant OS handles turning a VM into an execution environment. BareMetal and MirageOS follow this model. Why would you bother doing this instead of using Docker? That's a good question.
Indeed there is a JavaOS http://en.wikipedia.org/wiki/JavaOS
And here is discuss about why there is not many OS written in java Is it possible to make an operating system using java?
In short, Java need to run on JVM. JVM need to run on an OS. writing an OS using Java is not a good choice.
OS needs to deal with hardware which is not doable using java (except using JNI). And that is because JVM only provided limited commands which can be used in Java. These command including add, call a method and so on. But deal with hardware need command to operate reg, memory, CPU, hardware drivers directly. These are not supported directly in JVM so JNI is needed. That is back to the start - it is still needed to write an OS using C/assembly.
Hope this helps.
One of the main benefits of using Java is that abstracts away a lot of low level details that you usually don't really need to care about. It's those details which are required when you build an OS. So while you could work around this to write an OS in Java, it would have a lot of limitations, and you'd spend a lot of time fighting with the language and its initial design principles.
For operating systems you need to work really low-level. And that is a pain in Java. You do need e.g. unsigned data types, and Java only has signed data types. You need struct objects that have exactly the memory alignment the driver expects (and no object header like Java adds to every object).
Even key components of Java itself are no longer written in Java.
And this is -by no means- a temporary thing. More and more does get rewritten in native code to get better performance. The HotSpot VM adds "intrinsics" for performance critical native code, and there is work underway to reduce the overall cost of native calls.
For example JavaFX: The reason why it is much faster than AWT/Swing ever were is because it contains/uses a huge amount of native code. It relies on native code for rendering, and e.g. if you add the "webview" browser component it is actually using the webkit C library to provide the browser.
There is a number of things Java does really well. It is a nicely structured language with a fantastic toolchain. Python is much more compact to write, but its toolchain is a mess, e.g. refactoring tools are disappointing. And where Java shines is at optimizing polymorphism at run-time. Where C++ compilers would need to do expensive virtual calls - because at compile time it is not known which implementation will be used - there Hotspot can aggressively inline code to get better performance. But for operating systems, you do not need this much. You can afford to manually optimize call sites and inlining.
This answer does not mean to be exhaustive in any way, but I'd like to share my thoughts on the (very vast) topic.
Although it is theoretically possible to write some OS in pure java, there are practical matters that make this task really difficult. The main problem is that there is no (currently up to date and reliable) java compiler able to compile java to byte code. So there is no existing tool to make writing a whole OS from the ground up feasible in java, at least as far as my knowledge goes.
Java was designed to run in some implementation of the java virtual machine. There exist implementations for Windows, Mac, Linux, Android, etc. The design of the language is strongly based on the assumption that the JVM exists and will do some magic for you at runtime (think garbage collection, JIT compiler, reflection, etc.). This is most likely part of the reason why such a compiler does not exist: where would all these functionality go? Compiled down to byte code? It's possible but at this point I believe it would be difficult to do. Even Android, whose SDK is purely java based, runs Dalvik (a version of the JVM that supports a subset of the language) on a Linux Kernel.
If so, then how? I'm doing a team project for school. I thought that Java couldn't actually access hardware directly, since that would make it hard to be cross-platform. I need to know this, because after some quick Googling, I haven't found anything, and my team members(who want to do this and want to use Java) seem unsure of how to proceed- after apparently much more searching than I've done.
Your right in that you can't access hardware directly from Java (unless your calling on native code, but that's not what your after) since it runs in a sandboxed environment, namely the Java Virtual Machine (JVM).
However you can get some basic info from the JVM that it gathers from the underlying OS.
Take a look at using Java to get OS-level system information
What you are looking for is SIGAR API
Overview
The Sigar API provides a portable interface for gathering system information such as:
System memory, swap, cpu, load
average, uptime, logins
Per-process memory, cpu, credential
info, state, arguments, environment,
open files
File system detection and metrics
Network interface detection,
configuration info and metrics
TCP and UDP connection tables
Network route table
This information is available in most operating systems, but each OS has their own way(s) providing it.
SIGAR provides developers with one API to access this information regardless of the underlying platform.
The core API is implemented in pure C with bindings currently implemented for Java, Perl, Ruby, Python, Erlang, PHP and C#.
The amount of info you'll be able to get using native Java API's is pretty small, as your program generally only knows about the VM it's sitting on.
You can, however, call out to the command line, run native apps, and parse the results. It's not particularly good form in a Java app, however, as you lose the cross-platform benefits usually associated with the language.
You can get some basic information regarding the Processor/s using System.getEnv().
You can also use Runtime.getRuntime() - See the response of R. Kettelerij for details.
Another option is to use JMX. The MemoryMXBean for example provides some information regarding the RAM usage (heap and non-heap).
To anyone looking for a library that is still being maintained as of 2023 check out OSHI
Introduction
I heard something about writing device drivers in Java (heard as in "with my ears", not from the internet) and was wondering... I always thought device drivers operated on an operating system level and thus must be written in the same language as the OS (thus mostly C I suppose)
Questions
Am I generally wrong with this
assumption? (it seems so)
How can a driver in an "alien"
language be used in the OS?
What are the requirements (from a
programming language point of view)
for a device driver anyway?
Thanks for reading
There are a couple of ways this can be done.
First, code running at "OS level" does not need to be written in the same language as the OS. It merely has to be able to be linked together with OS code. Virtually all languages can interoperate with C, which is really all that's needed.
So language-wise, there is technically no problem. Java functions can call C functions, and C functions can call Java functions. And if the OS isn't written in C (let's say, for the sake of argument that it's written in C++), then the OS C++ code can call into some intermediate C code, which forwards to your Java, and vice versa. C is pretty much a lingua franca of programming.
Once a program has been compiled (to native code), its source language is no longer relevant. Assembler looks much the same regardless of which language the source code was written in before compilation. As long as you use the same calling convention as the OS, it's no problem.
A bigger problem is runtime support. Not a lot of software services are available in the OS. There usually is no Java virtual machine, for example. (There is no reason why there technically couldn't be, but usually, but usually, it's safe to assume that it's not present).
Unfortunately, in its "default" representation, as Java bytecode, a Java program requires a lot of infrastructure. It needs the Java VM to interpret and JIT the bytecode, and it needs the class library and so on.
But there are two ways around this:
Support Java in the kernel. This would be an unusual step, but it could be done.
Or compile your Java source code to a native format. A Java program doesn't have to be compiled to Java bytecode. You could compile it to x86 assembler. The same goes for whatever class libraries you use. Those too could be compiled all the way to assembler. Of course, parts of the Java class library requires certain OS features that won't be available, but then use of those classes could be avoided.
So yes, it can be done. But it's not straightforward, and it's unclear what you'd gain.
Of course another problem may be that Java won't let you access arbitrary memory locations, which would make a lot of hardware communication pretty tricky. But that could be worked around too, perhaps by calling into very simple C functions which simply return the relevant memory areas as arrays for Java to work on.
Writing Solaris Device Drivers in Java covers a A RAM disk device written in Java.
Another one for Linux. Goes more in depth on why you might want a DD in Java as well (since some people were wondering by the looks of the other posts and comments)
A device driver could be a lot of things
I actually write device drivers in java for a living: drivers for industrial devices, such as scales or weighing devices, packaging machines, barcode scanners, weighing bridges, bag and box printers, ... Java is a really good choice here.
Industrial devices are very different from your home/office devices (e.g. scanners, printers). Especially in manufacturing (e.g. food), companies opt more and more for a centralized server which runs an MES application (e.g. developed in Java) The MES server needs to interface with the devices of the production line, but also contains business logic. Java is a language that can do both.
Where your home/office devices are often built-in to your computer or connected with an USB cable, these industrial devices usually use Ethernet or RS232 connectors. So, in essence, pretty much every language could do the job.
There is not much standardisation in this area yet. Most vendors prefer to create their own protocol for their devices. After all they are hardware builders, not software geniuses. The result is that there is a high diversity of protocols. Some vendors prefer simple plain-text protocols, but others prefer complex binary protocols with CRC codes, framing, ... Sometimes they like to stack multiple protocols (e.g. a vendor specific handshaking algorithm on top of an OPC layer). A strong OOP language has a lot of advantages here.
E.g. I've seen java print at a continuous speed of 100ms/cycle. This includes generating a unique label, sending it to the printer, receiving a confirmation, printing it on paper and applying it to the product using air pressure.
In summary, the power of java:
It is useful for both business logic as complex interfacing.
It is just as reliable in communication with sockets as C.
Some drivers can benifit from Java's OOP power.
Java is fast enough.
It's not impossible, but possibly hard and possibly makes not much sense.
Possible is it, because Java is a normal programming language, as long as you have some way to access the data, it's no problem. Normally in a modern OS the kernel has a layer to allow raw access to hardware in some way. Also already exist drivers in userspace, at least the userspace-part should be no problem to implement in Java.
It makes possibly not too much sense, because the kernel has to start a JVM to execute the driver. Also JVM-implementations normally eat up much memory.
You could also use Java-code compiled to be executed natively on the platform (not with the help of a JVM). This is usually not that efficient, but it could be suitable for a device-driver.
The question is, does it make sense to implement the driver in Java? Or stated in another way: What is the benefit you hope for, if you use Java for implementing the driver instead of another alternative? If you can answer this question, you should find a way to make it possible.
At the end the hint to JNode, a project that tries to implement a complete OS purely based on Java.
You have a too narrow view of device drivers.
I have written such device drivers on top of MOST in an automotive application. A more widespread use might be drivers for USB devices if Java ever gets a decent USB library.
In these cases there is a generic low-level protocol which is handled in native code, and the Java driver handles the device specifics (data formats, state machines, ...).
For the motivation, please remember that there is plenty of fast languages which are better than C for programming; they might not be as fast as C, but they are safe languages: if you make a mistake you don't get undefined behavior. And "undefined behavior" includes executing arbitrary code supplied by some attacker which formats your HD.
Many functional languages are usually compiled to native code.
Device drivers contain the most bugs in an OS kernel - I know that for Linux (Linus Torvalds and others keep saying so) and I heard that for Windows. While for a disk or Ethernet driver you need top-notch performance, and while in Linux drivers today are the bottleneck for 10G Ethernet or SSD disks, most drivers don't need that much speed - all computers wait at the same speed.
That's why there are various projects to allow writing drivers which run outside of the kernel, even if that causes a slowdown; when you can do that, you can use whatever language you want; you will just then need Java bindings for the hardware control library you use - if you were writing the driver in C, you would still have a library with C bindings.
For drivers in kernel mode proper, there are two problems that I've not yet seen mentioned:
Garbage Collection, and that's a tough requirement. You need to write an in-kernel Garbage Collector; some GC algorithms rely on Virtual Memory, and you cannot use them. Moreover, you probably need to scan the whole OS memory to find roots for the GC. Finally, I would only trust an algorithm guaranteeing (soft) real-time GC, which would make the overhead even bigger.
Reading the paper which was mentioned about Java Device Drivers on top of Linux, they just give up, and require programmers to manually free memory. They try to argue that this will not compromise safety, but I don't think their argument is convincing - it's not even clear whether they understand that Garbage Collection is needed for a safe language.
Reflection and class loading. A full Java implementation, even when running native code, needs to be able to load new code. This is a library you can avoid, but if you have an interpreter or JIT compiler in kernel (and there's no real reason that makes it technically impossible).
Performance. The paper about a JVM on Linux is very bad, and their performance numbers are not convincing - indeed, they test a USB 1.1 network driver, and then show that performance is not so bad! However, given enough effort something better can surely be done.
Two last things:
I'd like to mention Singularity, which is a complete OS written in a C# variant, with just a Hardware Abstraction Layer in a native language.
About picoJava, it's a bad idea to use it unless your system is a really memory constrained one, like a smart card. Cliff Click already explained why: it gives better performance to write a good JIT, and nowadays even smartphones can support that.
Have you perhaps heard a reference to the JDDK?
Writing a device driver 100% in Java is not possible without native code to provide the interaction between (1) the OS-specific driver entry points and conventions, and (2) the JVM instance. The JVM instance could be started "in-process" (and "in-process" may have different meanings depending on the OS and on whether the driver is a kernel-mode or user-mode driver), or as a separate user-land process with which a thin, native driver adaptation layer can communicate and onto which the said driver adaptation layer can offload actual user-land work.
It is possible to compile java code to hardware native (i.e. not JVM bytecode) instructions. See for instance GCJ. With this in hand, you're a lot closer to being able to compile device drivers than you were before.
I don't know how practical it is, though.
Possible?
Yes but only in special circumstances. Because you can write an operating system in Java and C#, and then, should be able to write device drivers for it. The memory hit to these drivers and operating systems would be substantial.
Probable?
Not likely. Atleast not in the world of Windows or MacOS or even Linux... At least not anytime soon. Because languages like C# and Java depend on the CLR and JVM. The way these languages work means that they cannot effectively be loaded into ring0.
Also, the performance hit would be rather large if managed languages were employed in device drivers.
Device drivers have to be written in a language which can execute in the kernel, either compiled into it, or loaded as a module at runtime. This usually precludes writing device drivers in Java, but I suppose you theoretically could implement a JVM inside a device driver and let it execute Java code. Not that any sane person would want to do that.
On Linux there are several user-land (i.e. non-kernel) implementations of filesystems which uses a common abstraction layer called (fuse) which allows user-land programs to implement things which are typically done in the kernel.
The Windows Driver Foundation (WDF) is a Microsoft API that does allow both User and Kernel mode device drivers to be written. This is being done today, and it is now compatible with w2k and later (used to not have w2k as a supported target). There is no reason that JNI calls can't be made to do some work in the JRE . . . ( assuming that JNI is still the way to call Java from C/C++ . . . my knowledge is dated in that arena). This could be an interesting way to have high level algorithms directly munch on data from a USB pipe for something to that effect . . . cool stuff!
PCIe user space device drivers can be written in Pure Java. See JVerbs for details about memory-based direct hardware access, in the context of OFED. This is a technique that can be used to create very high performance systems.
You can examine the PCI bus to determine the memory regions for a given device, what ports it has, etc. The memory regions can be mapped into the JVM's process.
Of course, you're responsible for implementing everything yourself.
I didn't say easy. I said possible. ;)
See also Device Drivers in User Space, which discusses using the UIO framework to build a user space driver.
First of all, note that I'm not an expert on device drivers (though I wrote a few myself back in the day), much less an expert on Java.
Let's leave the fact that writing device drivers in a high-level language is not a good idea (for performance and possibly many other reasons) aside for a moment, and answer your question.
You can write device drivers in almost any language, at least in theory.
However, most device drivers need to do plenty of low-level stuff like handling interrupts and communicating with the OS using the OS APIs and system calls, which I believe you can't do in Java.
But, if your device communicates using, say, a serial port or USB, and if the OS doesn't necessarily need to be aware of the device (only your application will access the device*), then you can write the driver in any language that provides the necessary means to access the device.
So for example you probably can't write a SCSI card driver in Java, but you can write a driver for a proprietary control device, USB lava lamp, license dongle, etc.
* The obvious question here is, of course, does that count as a driver?