Postgres is now async by Default

Key Takeaways

So when we talk about uh postcris uh the worker method this is supported everywhere mac windows linux all platform that postcrist exist because just a worker it's just a process and that's the beauty of the worker right and that is the default so the IO method the new global user configuration gak they call it is now the default is worker and you're going to get the worker by default you so you're async by default and And I think they did it to flush out bugs, right? So you can switch that to make it IU ranking if you have your Linux a certain kernel version 5 whatever, right? And then you can have uh IO ring the postcus use ION to do the asynchronous IO's, right? So which is which is which is awesome. We talked about that. I talked about how I ring work. Essentially it's just um a similar idea. You sub there's a submission queue and there is a a what is it called the the receive queue not receive queue the the ready Q I think yeah so you submit your jobs here and then you get the out the the results from the other queue and then it's shared memory between the user space and the kernel space and that's basically where the all the security issues that happened few years ago without results from this shared memory between kernel it's it's a sin to share stuff between kernel and user space, right? We always have this kernel mode switch. We switch modes when we have when you switch from user to kernel. That's the only time you can read from kernel space. You cannot read as a user. You cannot read kernel stuff at all unless the you switch mode. So you can point to the kernel, you know, essentially code stack and read the kernel code to actually execute kernel stuff,