|
|
Subscribe / Log in / New account

Facebook and the kernel

Facebook and the kernel

Posted Mar 27, 2014 17:23 UTC (Thu) by dag- (guest, #30207)
In reply to: Facebook and the kernel by zblaxell
Parent article: Facebook and the kernel

I agree. Similar to nice and ionice, a way to disable caching for a (running) process would be very useful. A necessity really.

And it would also be useful for benchmarking purposes. See what the impact is for given workloads.


to post comments

Facebook and the kernel

Posted Mar 27, 2014 22:28 UTC (Thu) by giraffedata (guest, #1954) [Link]

a way to disable caching for a (running) process would be very useful.

I believe what the article refers to is something in between. No administrator has to hold the kernel's hand and tell it what to do, but no developer has to make an application give the kernel play by play descriptions of what the application is doing so the kernel knows what to do.

The middle ground is the application tells the kernel what it is generally doing, for example, "File X will normally be accessed in a log-type access pattern." The kernel figures out for itself that that means it should not waste cache space on data just written to the file, among other things.

Of course, that's not to say the direct adminstrator control wouldn't also be great, particularly for working around cases where the kernel doesn't yet have the required intelligence.


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds