What would a Minix 4 look like in a CPU everywhere world?

[DAN]

Hi,

since Minix is a research OS I was wondering if any OS researchers have given thought to the long-term design.

In our now cyber-physical, always connected, cloud services, block chained, IoT world how does the old OS model fit in moving forwards? It seems that there is plenty scope for research on small composable, flexible and adaptable OSs that can make use of computation cycles sitting anywhere. You could even say that botnets are an ingenious way to utilize distributed computation, even though that use is nefarious.

I imagine a small Minix core booting on a device. Then firing up any other cores present on the device. Then reaching out over the Internet to kick off processes on spare cores on remote CPUs (e.g. a home desktop or cloud systems, even GPUs). Processes then move around, sleep and awaken as cores get called upon for other tasks or connections die or are reborn. This is deeper than distributed mesh computing or grid computing. It is about the resilient, always on, updateable nature of Minix spread out over all of a person's devices or accesible computational resources. In an ideal world, any of the programs I run should be able to execute anywhere, moving around, with security built in, as computational demands require.

Was interested if anyone in the OS research field is considering such designs? Or put another way in the future OS world where would MINIX 4 head towards?

I'm not necessary after any answers here, but some pointers to resources in new thinking on future OS design would be useful.

DAN

[Jean-Baptiste Boric]

Hi,

What you are describing is basically SkyNet ;-)

MINIX is not a distributed operating system project and it is not meant to be one or become one. What you are looking for is more along the lines of Sprite, Amoeba, GLOBE and to a lesser extent Plan 9 from Bell Labs.

For MINIX4 itself, my opinion is that the micro-kernel has a good 30 years track record, but it's not really up to the task anymore. While dependable, the design is dated and the implementation is really showing its age, especially when compared to the competition (L4, HelenOS, QNX, INTEGRITY...). Lots of things like namespacing/containers, 64-bit support, SMP, kernel threads, VM hypervisor and so on currently can't be implemented at least partly because of it. I expect a hypothetical MINIX4 to either adopt another micro-kernel, or have a new design and implementation from scratch, and work up from there.