Messages in this thread | | | From | "Stefan Monnier" <monnier+lists/linux/kernel/news/@RUM.CS.YALE.EDU> | Subject | Re: elevator messages in 2.3.50 | Date | 9 Mar 2000 14:23:05 -0500 |
| |
>>>>> "Jamie" == Jamie Lokier <lk@tantalophile.demon.co.uk> writes: > `` Does a small holdoff time improve the average throughput? ''
I'm quite willing to believe that it does improve throughput if you have throughput to improve, as is the case on servers where several active processes are accessing the disk.
When there's only one active process, waiting is a waste of time and will just end up slowing you down, increasing your apparent disk latency.
So your idea sounds good up to a point, but we need some way to be able to decide "cleverly" whether to wait or not since waiting can be a very undesirable choice. Maybe we could wait only for requests coming from niced processes ?
Now that I think about it, that's maybe exactly what you were getting at. Then I completely agree: having some way to "slow-down" background I/O hogs would be great, since niceing those processes tends to be completely ineffective. If it turns out that nicing your squid/apache/foo server also improves its throughput, all the better!
Stefan
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |