I know this topic is already asked sometimes, and I have read almost all threads and comments, but I'm still not finding the answer to my problem.
I'm working on a high-performance network library that must have TCP server and client, has to be able to accept even 30000+ connections, and the throughput has to be as high as possible.
I know very well I have to use async methods, and I have already implemented all kinds of solutions that I have found and tested them.
In my benchmarking, only the minimal code was used to avoid any overhead in the scope, I have used profiling to minimize the CPU load, there is no more room for simple optimization, on the receiving socket the buffer data was always read, counted and discarded to avoid socket buffer fill completely.
The case is very simple, one TCP Socket listens on localhost, another TCP Socket connects to the listening socket (from the same program, on the same machine oc.), then one infinite loop starts to send 256kB sized packets with the client socket to the server socket.
A timer with 1000ms interval prints a byte counter from both sockets to the console to make the bandwidth visible then resets them for the next measurement.
I've realized the sweet-spot for packet size is 256kB and the socket's buffer size is 64kB to have the maximum throughput.
With the async/await type methods I could reach
~370MB/s (~3.2gbps) on Windows, ~680MB/s (~5.8gbps) on Linux with mono
With the BeginReceive/EndReceive/BeginSend/EndSend type methods I could reach
~580MB/s (~5.0gbps) on Windows, ~9GB/s (~77.3gbps) on Linux with mono
With the SocketAsyncEventArgs/ReceiveAsync/SendAsync type methods I could reach
~1.4GB/s (~12gbps) on Windows, ~1.1GB/s (~9.4gbps) on Linux with mono
Problems are the following:
async/await methods were the slowest, so I will not work with them
BeginReceive/EndReceive methods started new async thread together with the BeginAccept/EndAccept methods, under Linux/mono every new instance of the socket was extremely slow (when there was no more thread in the ThreadPool mono started up new threads, but to create 25 instance of connections did take about 5 mins, creating 50 connections was impossible (program just stopped doing anything after ~30 connections).
Changing the ThreadPool size did not help at all, and I would not change it (it was just a debug move)
The best solution so far is SocketAsyncEventArgs, and that makes the highest throughput on Windows, but in Linux/mono it is slower than the Windows, and it was the opposite before.
I've benchmarked both my Windows and Linux machine with iperf,
Windows machine produced ~1GB/s (~8.58gbps), Linux machine produced ~8.5GB/s (~73.0gbps)
The weird thing is iperf could make a weaker result than my application, but on Linux, it is much higher.
First of all, I would like to know if the results are normal, or can I get better results with a different solution?
If I decide to use the BeginReceive/EndReceive methods (they produced relatively the highest result on Linux/mono) then how can I fix the threading problem, to make the connection instance creating fast, and eliminate the stalled state after creating multiple instances?
I continue making further benchmarks and will share the results if there is any new.
================================= UPDATE ==================================
I promised code snippets, but after many hours of experimenting the overall code is kind of a mess, so I would just share my experience in case it can help someone.
I had to realize under Window 7 the loopback device is slow, could not get higher result than 1GB/s with iperf or NTttcp, only Windows 8 and newer versions have fast loopback, so I don't care anymore about Windows results until I can test on newer version. SIO_LOOPBACK_FAST_PATH should be enabled via Socket.IOControl, but it throws exception on Windows 7.
It turned out the most powerful solution is the Completion event based SocketAsyncEventArgs implementation both on Windows and Linux/Mono. Creating a few thousand instances of the clients never messed up the ThreadPool, the program did not stop suddenly as I mentioned above. This implementation is very nice to the threading.
Creating 10 connections to the listening socket and feeding data from 10 separate thread from the ThreadPool with the clients together could produce ~2GB/s data traffic on Windows, and ~6GB/s on Linux/Mono.
Increasing the client connection count did not improve the overall throughput, but the total traffic became distributed among the connections, this might be because the CPU load was 100% on all cores/threads even with 5, 10 or 200 clients.
I think overall performance is not bad, 100 clients could produce around ~500mbit/s traffic each. (Of course this is measured in local connections, real life scenario on network would be different.)
The only observation I would share: experimenting with both the Socket in/out buffer sizes and with the program read/write buffer sizes/loop cycles highly affected the performance and very differently on Windows and on Linux/Mono.
On Windows the best performance has been reached with 128kB socket-receive, 32kB socket-send, 16kB program-read and 64kB program-write buffers.
On Linux the previous settings produced very weak performance, but 512kB socket-receive and -send both, 256kB program-read and 128kB program-write buffer sizes worked the best.
Now my only problem is if I try create 10000 connecting sockets, after around 7005 it just stops creating the instances, does not throw any exceptions, and the program is running as there was no any problem, but I don't know how can it quit from a specific for loop without break, but it does.
Any help would be appreciated regarding anything I was talking about!
Because this question gets a lot of views I decided to post an "answer", but technically this isn't an answer, but my final conclusion for now, so I will mark it as answer.
About the approaches:
The async/await functions tend to produce awaitable async Tasks assigned to the TaskScheduler of the dotnet runtime, so having thousands of simultaneous connections, therefore thousands or reading/writing operations will start up thousands of Tasks. As far as I know this creates thousands of StateMachines stored in ram and countless context switchings in the threads they are assigned to, resulting in very high CPU overhead. With a few connections/async calls it is better balanced, but as the awaitable Task count grows it gets slow exponentially.
The BeginReceive/EndReceive/BeginSend/EndSend socket methods are technically async methods with no awaitable Tasks, but with callbacks on the end of the call, which actually optimizes more the multithreading, but still the limitation of the dotnet design of these socket methods are poor in my opinion, but for simple solutions (or limited count of connections) it is the way to go.
The SocketAsyncEventArgs/ReceiveAsync/SendAsync type of socket implementation is the best on Windows for a reason. It utilizes the Windows IOCP in the background to achieve the fastest async socket calls and use the Overlapped I/O and a special socket mode. This solution is the "simplest" and fastest under Windows. But under mono/linux, it never will be that fast, because mono emulates the Windows IOCP by using linux epoll, which actually is much faster than IOCP, but it has to emulate the IOCP to achieve dotnet compatibility, this causes some overhead.
About buffer sizes:
There are countless ways to handle data on sockets. Reading is straightforward, data arrives, You know the length of it, You just copy bytes from the socket buffer to Your application and process it.
Sending data is a bit different.
You can pass Your complete data to the socket and it will cut it to chunks, copy the chucks to the socket buffer until there is no more to send and the sending method of the socket will return when all data is sent (or when error happens).
You can take Your data, cut it to chunks and call the socket send method with a chunk, and when it returns then send the next chunk until there is no more.
In any cases You should consider what socket buffer size You should choose. If You are sending large amount of data, then the bigger the buffer is, the less chunks has to be sent, therefore less calls in Your (or in the socket's internal) loop has to be called, less memory copy, less overhead.
But allocating large socket buffers and program data buffers will result in large memory usage, especially if You are having thousands of connections, and allocating (and freeing up) large memory multiple times is always expensive.
On sending side 1-2-4-8kB socket buffer size is ideal for most cases, but if You are preparing to send large files (over few MB) regularly then 16-32-64kB buffer size is the way to go. Over 64kB there is usually no point to go.
But this has only advantage if the receiver side has relatively large receiving buffers too.
Usually over the internet connections (not local network) no point to get over 32kB, even 16kB is ideal.
Going under 4-8kB can result in exponentially incremented call count in the reading/writing loop, causing large CPU load and slow data processing in the application.
Go under 4kB only if You know Your messages will usually be smaller than 4kB, or just very rarely over 4KB.
My conclusion:
Regarding my experiments built-in socket class/methods/solutions in dotnet are OK, but not efficient at all. My simple linux C test programs using non-blocking sockets could overperform the fastest and "high-performance" solution of dotnet sockets (SocketAsyncEventArgs).
This does not mean it is impossible to have fast socket programming in dotnet, but under Windows I had to make my own implementation of Windows IOCP by directly communicating with the Windows Kernel via InteropServices/Marshaling, directly calling Winsock2 methods, using a lot of unsafe codes to pass the context structs of my connections as pointers between my classes/calls, creating my own ThreadPool, creating IO event handler threads, creating my own TaskScheduler to limit the count of simultaneous async calls to avoid pointlessly much context switches.
This was a lot of job with a lot of research, experiment, and testing. If You want to do it on Your own, do it only if You really think it worth it. Mixing unsafe/unmanage code with managed code is a pain in the ass, but the end it worth it, because with this solution I could reach with my own http server about 36000 http request/sec on a 1gbit lan, on Windows 7, with an i7 4790.
This is such a high performance that I never could reach with dotnet built-in sockets.
When running my dotnet server on an i9 7900X on Windows 10, connected to a 4c/8t Intel Atom NAS on Linux, via 10gbit lan, I can use the complete bandwidth (therefore copying data with 1GB/s) no matter if I have only 1 or 10000 simultaneous connections.
My socket library also detects if the code is running on linux, and then instead of Windows IOCP (obviously) it is using linux kernel calls via InteropServices/Marshalling to create, use sockets, and handle the socket events directly with linux epoll, managed to max out the performance of the test machines.
Design tip:
As it turned out it is difficult to design a networking library from scatch, especially one, that is likely very universal for all purposes. You have to design it to have many settings, or especially to the task You need.
This means finding the proper socket buffer sizes, the I/O processing thread count, the Worker thread count, the allowed async task count, these all has to be tuned to the machine the application running on and to the connection count, and data type You want to transfer through the network. This is why the built-in sockets are not performing that good, because they must be universal, and they do not let You set these parameters.
In my case assingning more than 2 dedicated threads to I/O event processing actually makes the overall performance worse, because using only 2 RSS Queues, and causing more context switching than what is ideal.
Choosing wrong buffer sizes will result in performance loss.
Always benchmark different implementations for the simulated task You need to find out which solution or setting is the best.
Different settings may produce different performance results on different machines and/or operating systems!
Mono vs Dotnet Core:
Since I've programmed my socket library in a FW/Core compatible way I could test them under linux with mono, and with core native compilation. Most interestingly I could not observe any remarkable performance differences, both were fast, but of course leaving mono and compiling in core should be the way to go.
Bonus performance tip:
If Your network card is capable of RSS (Receive Side Scaling) then enable it in Windows in the network device settings in the advanced properties, and set the RSS Queue from 1 to as high you can/as high is the best for your performance.
If it is supported by Your network card then it is usually set to 1, this assigns the network event to process only by one CPU core by the kernel. If You can increment this queue count to higher numbers then it will distribute the network events between more CPU cores, and will result in much better performance.
In linux it is also possible to set this up, but in different ways, better to search for Your linux distro/lan driver information.
I hope my experience will help some of You!
I had the same problem. You should take a look into:
NetCoreServer
Every thread in the .NET clr threadpool can handle one task at one time. So to handle more async connects/reads etc., you have to change the threadpool size by using:
ThreadPool.SetMinThreads(Int32, Int32)
Using EAP (event based asynchronous pattern) is the way to go on Windows. I would use it on Linux too because of the problems you mentioned and take the performance plunge.
The best would be io completion ports on Windows, but they are not portable.
PS: when it comes to serialize objects, you are highly encouraged to use protobuf-net. It binary serializes objects up to 10x times faster than the .NET binary serializer and saves a little space too!
Is there any way (preferably in C#) how to regularly measure connection layer latency (roundtrip) without changing the application protocol and without creating separate dedicated connection - e.g. using some similar SYN-ACK trick like tcping do but without closing/opening connection?
I'm connecting to the servers via given ASCII based protocol (and always using TCP_NODELAY). Servers send me large amount of discrete messages and I'm regularly sending 'heartbeat' payload (but there is no response payload to the heartbeat).
I cannot change the protocol and in many cases I also cannot create more than one physical connection to the server.
Keep in mind that TCP does windowing, so this could cause issues when trying to implement an elegant SEQ/ACK solution. (you would want sequence, not synchronize)
[EDIT: Snipped a very overcomplicated and confusing explaination.]
I'd have to say the best way is to use a simple stopwatch method of starting a timer, making a very thin request or poll, and measure the time back from it. If that query really is the lightest you can make it, then that should give you the minimum amount of time you can reasonably expect to wait, which sometimes more valuable than the ping (which can be misleading).
If you really absolutely need just the network time to machine and back, just use an ICMP ping.
I developed a simple UDP message server and client application in Windows, the server can send a message to the client but the client can't send anything, they are only listening. the problem is the client application is use quite big memory usage is about 7M when it's listening and 9M when it received a packet. Could I reduce the memory usage into at least less then 1M?
How are you measureing your memory footprint? Any managed .net application, even the smallest typically has a shared working set of around 50 MBs, the actual memory footprint of your app is much smaller than that.
Have you tried calling GC.GetTotalMemory to look at the actual managed memory usage?
Much of this burden is the overhead of running the whole CLR system, garbage collection etc. If you're super-sensitive to memory footprint (<10-20Mb) then the CLR may not be for you. Even a basic HelloWorld private working set is over 4Mb of RAM.
If you are sensitive to footprint, you might be best served by looking to a true ahead-of-time compiled language like C/C++ etc.
I'm building a network server and starting a lot of AppDomains on the server to which requests are routed. What will be the fastest way to send off a request payload to one of the AppDomains for processing?
Read in the payload from the socket into a byte array and marshal it.
Marshal the network stream (inherits from MarshalByRef) to the AppDomain.
Read the payload. Decode it into objects. Marshal the decoded objects.
Use named pipes to transfer the byte array.
Use loopback sockets.
Maybe there is a way to marshal the actual socket connection?
The decoding mostly creates immutable objects that are used to determine how to fulfill the clients request and the AppDomain then creates a response and marshals it back to the host AppDomain which sends it back through the socket.
The method should prefer less memory over less CPU.
WCF is not an option.
TCP binary remoting is certainly fast, I do not how much faster it is than raw sockets which is probably the fastest, but a royal PIA.
I have run 1500 - 2000 req per second in production using HTTP binary remoting between two boxes. On the same box you should have much high performance using TCP or a name pipes channel, depending in the CPU cycles it takes to process the data.
If I was you I would take a look at how Cassini is implemented. It does pretty much exactly what you are talking about doing.
Actually Cassini has been sort of superceded by Webhost which is the built-in webserver that ships with Visual Studio now. Take a look at this post on Phil Haack's blog for more.
Very good question. If I were coming at this problem I would probably use a Buffered Stream / Memory Stream and marshal the stream into the AppDomain that consumes the object to reduce marshaling or serializing many object graphs that were created in a different AppDomain.
But then again, it sounds like you are almost completely duplicating the functionality of IIS, so I would look/reflector into the System.Web.Hosting namespace and see how they handle it and their WorkerThreadPool etc....
6 .Maybe there is a way to marshal the
actual socket connection?
6-th is IMO the best option.
Socket from process perspective is just a handle. AppDomains reside in single process. That means that appdomains can interchange socket handles.
If socket marshalling is not working, you can try recreating socket in other appdomain. You can use DuplicateAndClose to do this.
If that will not work, you should do some perfomance testing to choose the best data transfer method. (I would choose named pipes or memomry mapped files)
We need to develop some kind of buffer management for an application we are developing using C#.
Essentially, the application receives messages from devices as and when they come in (there could be many in a short space of time). We need to queue them up in some kind of buffer pool so that we can process them in a managed fashion.
We were thinking of allocating a block of memory in 256 byte chunks (all messages are less than that) and then using buffer pool management to have a pool of available buffers that can be used for incoming messages and a pool of buffers ready to be processed.
So the flow would be "Get a buffer" (process it) "Release buffer" or "Leave it in the pool". We would also need to know when the buffer was filling up.
Potentially, we would also need a way to "peek" into the buffers to see what the highest priority buffer in the pool is rather than always getting the next buffer.
Is there already support for this in .NET or is there some open source code that we could use?
C# sharps memory management is actually quite good, so instead of having a pool of buffers, you could just allocate exactly what you need and stick it into a queue. Once you are done with buffer just let the garbage collector handle it.
One other option (knowing only very little about your application), is to process the messages minimally as you get them, and turn them into full fledged objects (with priorities and all), then your queue could prioritize them just by investigating the correct set of attributes or methods.
If your messages come in too fast even for minimal processing you could have a two queue system. One is just a queue of unprocessed buffers, and the next queue is the queue of message objects built from the buffers.
I hope this helps.
#grieve: Networking is native, meaning that when buffers are used the receive/send data on the network, they are pinned in memory. see my comments below for elaboration.
Why wouldn't you just receive the messages, create a DeviceMessage (for lack of a better name) object, and put that object into a Queue ? If the prioritization is important, implement a PriorityQueue class that handles that automatically (by placing the DeviceMessage objects in priority order as they're inserted into the queue). Seems like a more OO approach, and would simplify maintenance over time with regards to the prioritization.
I know this is an old post, but I think you should take a look at the memory pool implemented in the ILNumerics project. I think they did exactly what you need and it is a very nice piece of code.
Download the code at http://ilnumerics.net/ and take a look at the file ILMemoryPool.cs
I'm doing something similar. I have messages coming in on MTA threads that need to be serviced on STA threads.
I used a BlockingCollection (part of the parallel fx extensions) that is monitored by several STA threads (configurable, but defaults to xr * the number of cores). Each thread tries to pop a message off the queue. They either time out and try again or successfully pop a message off and service it.
I've got it wired with perfmon counters to keep track of idle time, job lengths, incoming messages, etc, which can be used to tweak the queue's settings.
You'd have to implement a custom collection, or perhaps extend BC, to implement queue item priorities.
One of the reasons why I implemented it this way is that, as I understand it, queueing theory generally favors single-line, multiple-servers (why do I feel like I'm going to catch crap about that?).