>> Coroutines based on stack manipulation(as opposed to simple FSMs) are extremely powerful and are particularly useful for implementing fair scheduling across runnable ‘processes’ and concurrent execution contexts.

So, isn't this what threads are for? This is not a snarky comment, genuine interest: but isn't this what the OS scheduler and thread implementations are meant to do, and implementing essentially ones own threads and scheduling is the epitomy of 'reinventing the wheel'? Specifically, when going down the route of using a coroutine to essentially duplicate functionality provided by native threads, is there any performance advantage? My understanding is that, nowadays, OS threads and schedulers are very highly optimize for performance.

Other than being able to write concurrent code with yield() and resume(), is there really any reason to NOT use OS level primitives?

Operating system threads have two big disadvantages. You also design the code differently, since one schedule is cooperative and the other is preemptive.

First, OS context switching is much more expensive in terms of overhead compared to lightweight contexts like fibers. A highly optimized user-controlled context-switching scheme can be barely more expensive than a function call. Context switching OS threads, even when highly optimized, is much more expensive than you might assume, microseconds compared to a few dozen clock cycles, which adds up if you need to switch contexts frequently. In many software systems, the computational cost of the OS context switch can be higher than the computational cost of the real work done during that time slice (i.e. >50% CPU overhead due to context switching) and you end up doing a lot of optimization work trying to avoid this situation, often by implementing something that looks suspiciously like user-space coroutines.

Second, and I would argue more importantly, the operating system provided thread schedule, through no fault of its own, is often pretty terrible for concurrent/distributed/parallel systems. The operating system simply lacks the runtime context to optimally determine (1) when to switch contexts and (2) the right thread to switch to. In practice, this leads to a lot more context switching, and wasteful context switching, with its high overhead than would ever occur if you designed your own schedule. The impact is not small at all, a poor schedule can seriously damage CPU cache locality.

Linux threads are impressively optimized for what they do. The way to beat them is to not do as much. Specifically, to not do a kernel transition. Many coroutine switch implementations are not much more expensive than function calls. For example https://news.ycombinator.com/item?id=13199581

That's a good question, and the answer has to do with, obviously, the cost of spawning OS threads/processes, context-switching among them, and how all that affects caches (cache hit/miss ratio dominates most workloads nowadays, more so than actual CPU instructions execution). Context switching can be cost on average 30micros of CPU overhead, but can reach as high as 1500micros. (You can reduce that by pinning threads to core, but the savings will likely get may be as much as 20% - maybe). Thankfully now, at least on Linux, AFAIK, a context switch doesn't result in a TLB flush - back in the day that was quite expensive. Whenever the OS scheduler needs to context switch, it needs to save old state(including regs), and restore another state and perform some other kind of book-keeping. This is more expensive and convoluted than doing it yourself -- you also usually don't need to care for all possible registers, so you can reducing switching costs by e.g not caring about SSE registers. Apps that create too many threads are constantly fighting for CPU time - you will end up wasting a lot of cpu cycles practically cycling between threads.

Another hidden cost that can severely impact server workloads is that, after your thread has been switched out, even if another process becomes runnable, it will need to wait in the kernel's run queue until a CPU core is available for it. Linux kernels are often compiled with HZ=100, which means that processes are given time slices of 10ms. If your thread has been switched out, but becomes runnable almost immediately, and there are 2 thread before your threads before it in the run queue waiting for CPU time, your thread may have to wait upto 20ms in a worse case scenario to get CPU time. Depending on the average length of the run queue (reflected in sys.load average), and how length threads typically run before getting switched out again, this can considerable affect performance.

Also, kernels don't generally do a good job at respecting CPU affinity -- even on an idle system. You can control that yourself.

If you are going to represent ‘jobs’ or tasks as threads, and you need potentially 100s or 1000s of those, it just won’t scale - the overhead is too high.

If you control context-switching on your application (in practice that’s about saving and restoring a few registers) -- cooperative multitasking -- you can have a lot more control on how that works, when its appropriate to switch etc, and the overhead is far lower.

Oh, and on average its 2.5-3x more expensive to context switch when using virtualisation.

>> If you are going to represent ‘jobs’ or tasks as threads, and you need potentially 100s or 1000s of those, it just won’t scale - the overhead is too high.

This seems to be an indictment against containerization (ie docker, kubernetes, etc). How can a single machine, even with containers, scale out if this is true? Do people advocating docker and similar technologies not understand or recognize that their containers should only utilize y threads (where y = (total threads per machine)/(number of containers hosted on that machine))? I personally have never heard this mentioned as a consideration when deploying docker, for example.

The advantages of containers are multidimensional. The benefit of running multiple processes on a single node is not solely to increase utilization of a given machine but to also gain advantages in deployment, etc.

Schedulers can take what's being mentioned here into account. You can say "I need at least 2 cores but would like up to 8" or "I need guaranteed 8 cores, evict lower priority containers if necessary" to ensure your containers aren't fighting over resources . You can also classify machines for certain types of workloads and then specify certain containers should run on a particular class of node. Not all container platforms handle this, but Kubernetes does at least.

And as we add improved resource management and tuning to Kubernetes you gain those improvements without having to change your deployment pipeline - just like when a compiler gets better you get more performance for "free".