But I really wanted some explanation of why Windows process startup seems to be so heavyweight. Why does anything that spawns lots of little independent processes take so bloody long on Windows?

I'm not saying "lots of processes on Windows is slow, lots of processes on Linux is fast, Windows uses CreateProcess, Linux uses fork, CreateProcess is an alternative to fork/exec, therefore fork/exec is better than any alternative." I can imagine all kinds of reasons for the observed behavior, few of which would prove that fork is a good model. But I still want to know what's going on.

Many frameworks are backed by XPC services, where the parent process has a socket-like connection to a backend server. After forking, the child would have no valid connection to the server. The fork() function establishes a new connection in the child for libSystem, to allow Unix programs to port easily to macOS, but other services' connections are not re-established. This makes fork on macOS (i) slow, and (ii) unsafe for code that touches virtually any of Apple's APIs.

Fork() is now basically the root of a looong list of special cases in so many aspects of programming. Things get even worse when you use a language with built-in runtime such as Golang for which multi-threaded programming the default behaviour. If fork() can't even handle multiple threads, what is the real point of having it when a 8 core 16 threads AMD processor is about $150 each.

The paper also mention the use case of multiprocess servers which relies heavily on fork() but dismiss it as it could be implemented with threads. A crash in a worker would lead to the crash of the whole application. While a worker could just be restarted.

A proper use case of removing fork() from an actual program would help. For example, how nginx on Windows is implemented?

In section 7 it suggests "We should therefore strongly discourage the use of fork in new code, and seek to remove it from existing apps."

Is anyone here going to help work on changing those 1304 packages?

I have already over-volunteered for thankless FOSS tasks like this, so I know it won't be me.

> 7. GET THE FORK OUT OF MY OS!

Someone couldn't resist...

Interested to see what this paper has to say.

Fork-requiring program class 1:

The biggest example where fork() is needed are webservers/long-running programs with significant unchanging memory overhead and/or startup time.

Many large applications written in a language or framework that prefers the single-process/single-thread model for executing requests (e.g. Python/gunicorn, Perl, a lot of Ruby, NodeJS with ‘cluster’ for multicore, etc.) are basically dependent on fork(). Such applications often have a huge amount of memory required at startup (due to loading libraries and initializing frameworks/constant state). Creating workers that can execute requests in parallel but don’t require any additional memory overhead (just what they consume per request) is essential for them. fork()ing without exec()ing a new program facilitates this memory sharing; everything is copy-on-write, and most big webapps don’t need to write most of the startup-initialized memory they have, though they may need to read it.

Additionally, starting up such programs can take a long time due to costly initialization (seconds or minutes in the worst cases); using fork() allows them to quickly replace failed or aged-out subprocesses without having to pay that overhead (which also typically pegs a CPU core) to change their parallelism. “Quickly” might not be quick enough if a program needs to continually launch new subprocesses, but for periodically forking (or just forking-at-startup) long-running servers with a big footprint, it’s far better than re-initializing the whole runtime. For better or worse, we’ve come far enough from old-school process-per-request CGI that it is no longer feasible in most production deployments.

Anticipated rebuttals:

Q: Wouldn't it be nice if everyone wrote apps small enough that startup time was minimized and memory footprint was low?

A: Sure, but they won’t.

Q: People should just write their big, long-running services in a framework that starts fast, has low memory requirements, and uses threads instead of fork()s.

A: See previous answer. Also see zzzcpan’s response.

Q: Can you access some of those benefits with careful use of shared memory?

A: Yes, but it’s much harder to do than it is to use fork() in most cases (caveat Windows, but it’s still hard).

Q: Do tools exist in single-proc/single-thread forking frameworks/languages which switch from forking to hybrid async/threaded paradigms (like gevent) instead?

A: Yes, but they’re not nearly as mature, capable, or useful (especially when you need to utilize multiple cores).

Fork-requiring program class 2:

Programs which fork infrequently in order to parallelize uncommon tasks over shared memory. Redis does this to great effect; it doesn’t exec(), it just forks off a child process which keeps the memory image at the time of fork from the parent, and writes most of that memory state to disk so that the parent can keep handling requests while the child snapshots.

Python’s multiprocessing excels at these kinds of cases as well. If you’re launching and destroying multiprocessing pools multiple times a second, then sure, you’re holding it wrong, but many people get huge wins from using multiprocessing to do parallel operations on big data sets that were present in memory at the time multiprocessing fork()ed off processes. While this isn’t cross-platform, it can be a really massive performance advantage: no need to serialize data and pass it to a multiprocessing child (this is what apply_async does under the covers) if the data is already accessible in memory when the child starts. Node's 'cluster' module will do this too, if you ask nicely. Many other languages and frameworks support similar patterns: the common thread is making fork()ing parallelism "easy enough" with the option of spending a little extra effort to make it really really cheap to get pre-fork memory state into children for processing. Oh, and you basically don't have to worry about corrupting anyone else's in-memory state if you do this (not so with threads).

Anticipated Rebuttals:

Q: $language provides a really accessible way to use true threads that isn’t nearly as tricky as e.g. multiprocessing or knowing all the gotchas (e.g. accidental file descriptor sharing between non-fork-safe libraries) of fork(); why not use that?

A: Many people still prefer languages with primarily-forking parallelism[1] constructs for reasons besides their fork-based concurrency capabilities--nobody’s claiming multiprocessing beats goroutines for API friendliness--so fork() remains useful in much more than a legacy capacity.

Q: Why not use $tool which does this via threads or why not bind $threaded_language to $scripting_language and use threads on the other side of the FFI boundary?

A: People won’t switch. They won’t switch because it’s hard (don't tell me threaded Rust is as easy to pick up as multiprocessing--Rust has a lot of advantages in this space, but that ain't one of them) and because there’s a positive benefit to staying within a given platform, even if some infrequent tasks (hopefully your Python doesn’t invoke multiprocessing too much) are a bit more cumbersome than usual. Also, “Friendly, easy-to-use concurrency with threads” is often a very false promise. There’s a reason Antirez is resistant to threading.

--------------

TL;DR perhaps using fork() and exec() for launching new programs needs to stop. But fork() itself is absolutely essential for common real-world use cases.

[1] References to parallelism via fork() above assume you have more than one core to schedule processes onto. Otherwise it’s not that parallel.

EDITs: grammar. There will be several because essay. I won't change the substance.

However, may I point out that Microsoft SQL Server benchmarks have been posted that show Linux TCP-H outperforming Windows?

https://www.dbbest.com/blog/running-sql-server-on-linux/

While I am sure that this is wise criticism, it might also be concluded that Windows itself contains no small amount of architectural decisions that limit performance.

The use of fork as a concurrency mechanism (creating a new thread of control that executes in a copy of the address space) is very good and useful.

In the POSIX shell language, the subshell syntax (command1; command2; ...) is easily implemented using fork. This is useful: all destructive manipulations in the subshell like assignments to variables or changing the current directory do not affect the parent.

Check out the fork-based Perl solution to the Amb task in Rosetta code: https://rosettacode.org/wiki/Amb#Using_fork

This essentially simulates continuations (in a way). (If the parent process does nothing but wait for the child to finish, fork can be used to perform speculative execution, similar to creating a continuation and immediately invoking it).

Microsoft "researchers" can stuff it and their company's flagship piece of shit OS.

also:

> When a fork syscall is made on WSL, lxss.sys does some of the initial work to prepare for copying the process. It then calls internal NT APIs to create the process with the correct semantics and create a thread in the process with an identical register context. Finally, it does some additional work to complete copying the process and resumes the new process so it can begin executing.

https://blogs.msdn.microsoft.com/wsl/2016/06/08/wsl-system-c...

Methinks MS need to focus on their own issues and leave the nix world alone. While many people find their involvement in FOSS welcome, I do not and never have. They are still a for-profit company beholden to shareholders.

The purchase by MS of GitHub may, again, be welcomed by many, but I find it disastrous. I smell triple E here no matter what anyone says. This is why distros like Debian and Slackware are still so important. All nix needs to do is start adopting MS ideas and then it's a matter of time before distros adopt disastrous code like systemd. MS does want to control everything around them like every other for-profit company. I cannot see this any other way. They are involved for their own good, for things like Azure and their own "cloud". MS needs to focus on their own garden and not that of *nix. I always have and always will prefer the "us and them" mentality when dealing with MS. Don't forget EEE. It's still a reality should you care to look hard enough.

    * redirect stdin, stdout, and stderr
    * open files that might be needed and close files that aren't
    * change process limits
    * drop privileges 
    * change the root directory
    * change namespaces

Regardless of this paper, I don't see its use declining significantly any time soon.

Another common use of fork() for things other than exec()ing is multi-process services where all will keep running te same program. Arranging to spawn or vfork-then-exec self and have the child realize it's a worker and not a (re)starter is more work because a bunch of state needs to be passed to the child somehow (via an internal interface), and that feels hackish... And also this case doesn't suffer much from fork()s badness: you fork() early and have little or no state in the parent that could have fork-unsafety issues. But it's worth switching this use-case to spawn or vfork-then-exec just so we have no use cases for fork() left.

I don't think it is suboptimal. As the paper acknowledges it primary use is to set up the environment of the program you are about to exec(). There are four points to be made about that:

1. If you don't need to set up the environment it imposes almost no coding overhead. It reduces to "if (!(pid = fork()) exec(...)". That's hardly a huge imposition.

2. It doesn't seem to impose much runtime overhead either. If it did Linux and BSD would have acquired a spawn() syscall's ages ago. As it is they all implement posix_spawn() using a vfork() / exec(). Given we are talking a 30 year history here any claims getting rid of the fork() would give a noticeable performance boost should not be taken seriously without evidence.

3. If you do need to setup the environment then yes there are traps with threads and other things. As the paper says it's terrible - but to paraphrase Churchill the one thing it has in it's favour is it's better than all the other ways of doing the same thing. They actually acknowledge how to replace flexibility allowed by fork() is an open research question. "We think it's horrible, but we don't have an alternative" isn't a convincing argument.

4. For all it's faults fork() has one outstanding attribute - it's conceptually drop dead simple: "create an exact copy of the process, the sole difference being getpid() returns a different value". That translates to bugger all code needed to implement it, few bugs, small man pages and a simple interface. A replacement providing the same flexibility will be some hideously complex thing that tries to implement all the use cases people used fork() for. It will be big and hard to learn, hard to use correctly, take reams of code, still won't do all that fork() allowed you to do. We will be complaining about if for decades to come.

I stopped reading the paper when they claims O_CLOEXEC was an overhead imposed by fork(). It isn't. The telltale give away should be it doesn't take effect on a fork() - it happens on the exec(), and the spawn() or whatever does exec()'s job. If you remove fork() things like O_CLOEXEC is your only way to control what environment your child process gets. Therefore one outcome of removing fork() is the reverse of what they claim - you won't get less O_CLOEXEC's, you will get many, many more of them as programmers clamour for ways to do the things fork() allowed them to do.

They also point out that on modern hardware you often should want to write multithreaded multiprocess application.

Their main criticism of fork is that it does not compose at any level of the OS (as it cannot be implemented over a different primitive)

I understand that a lot of people here dislike Microsoft for good reason (not only historical), but drawbacks in fork() are well known and recognized, here they point out that it is also hard-to-impossible to implement as a compatibility layer if the kernel does not support fork.

Also:

> Microsoft "researchers" can stuff it and their company's flagship piece of shit OS.

Do you have any reason to insult Microsoft researchers? They have plenty of citations in this paper of other researchers that appear to agree with them. This type of comments does not appear constructive to me

The paper mentions the benefit of posix_spawn for the fork+exec use case.

I might've seen posix_spawn while skimming a manpage or browsing a change log but this is the first time that I'd actually learned about its purpose.

The article's conclusion isn't "and therefore Linux is bad" btw.

Of course, all Windows APIs are terrible, but that doesn't make complaints about fork() any less legitimate. The concept of Establishing empty processes, instead of cloning yourself, is much more sane.

After all, the use of fork() is 99% of the time just to call execve(), and anything done in between is just to clean up the mess from fork(). Having a dedicated way to just create processes in a controlled fashion would have been better there. And, the other 1% is usually cases where pthread should have been used instead.

I don't run Windows so I'm far from the most biased person but frankly, on the surface the fork/exec thing really does seem unnecessary and weird in the modern world, where we've come up with better ways to do concurrency than just raw threads and processes anyways.

Win32 API cannot do it. The underlying NT kernel can.

Their arguments of why fork() is not a good fit these days seemed pretty reasonable to me.