[FoRK] [info] (highscalability.com) The Secret to 10 Million Concurrent Connections -The Kernel is the Problem, Not the Solution

Stephen Williams sdw at lig.net
Mon May 20 10:01:47 PDT 2013


On 5/20/13 9:54 AM, Tomasz Rola wrote:
> On Mon, 20 May 2013, Stephen Williams wrote:
>
>> As someone on page comments pointed out, this has been thought of and done
>> before:
>> http://pdos.csail.mit.edu/exo.html
> Right, it had been done, or at least thought of and maybe prototyped
> (because they hopefully based their papers on running code experiences)
> before.
>
> I'm not going to diminish their work, but the page is 15 years old and
> their newest paper on exokernel is 11 years old. If anything came out of
> this effort, I don't think I heard about it. Which of course has nothing
> to do with the quality of their concepts but rather.

We've all been waiting for an implementation, so hopefully they will get 
it done.
We've all known that there was plenty of efficiency to be had in this 
direction.

> ...
>
> To me, the question that hit me was, what kind of stuff I could do if I
> could handle 10MC? Let's put technical difficulties aside for a while.
>
> BTW, article also gives some arguments to those who would like to have
> some and try breaking out of psychological addiction to monoblock kernel.
> Modules or no modules, it still smells of monoblock to me (I speak of
> Linux only). Perhaps Hurd will see some new developers. Or maybe even
> something better will appear.

Repeatedly debunked, but you're welcome to try again.  Kernel threads, 
queues, etc., and clever driver handling, can probably accomplish the 
same things more efficiently.

>
> [...]
>> Let me know when they have something useful actually running.
> Let's wait and see. Who knows, maybe a year or two.
>
> Regards,
> Tomasz Rola
>
>

sdw



More information about the FoRK mailing list