Messages in this thread | | | Subject | Re: atomicity | Date | Tue, 8 Dec 1998 12:49:23 +0100 (MET) | From | (Rogier Wolff) |
| |
Alan Cox wrote: > > I am astonished to hear that ext2fs uses cyl groups. > > They still make sense
And they are no longer "cyl groups", but they are block groups.
Assuming a 1k block size, the block groups are 8Mb. No attempt is made to align them to cylinder boundaries.
> > not useful anymore. With modern disks, chances are excellent that the > > specific 3D geometry the file system sees is an utter fiction, and > > that cylinder locality is thus more-or-less a meaningless concept. > > If it was meaningless then things like sorting write queues would also > be pointless. Its certainly true that the 3D information is less than > useful nowdays. Its not however true that relative distance (in time) > between sectors is not strongly proportional to the difference in sector > numbers, or that reading sectors is order of sequentional logical number > isnt a good thing
REW: E-too-many-negatives: Brain overflow.
Reading a block that is 2 blocks away from the last block you read will just take (2 blocks / raw_disk_speed). That's very proportional to the difference in sector numbers. But it is a "law" that only holds for "small" values of the difference. (It holds upto about 10ms: the time it takes for a full rotation)
So, once you leave the current track, you're usually facing a seek time (or head-switch-time (*)) of about 3ms, plus the average rotational latency of about 5ms. A "long distance" seek is not that much worse: as a disk claims an average seek time of 8 or 10 ms, which is only about three times the track-to-track seek time.
Roger.
-- My pet light bulb is a year old today. \_________ R.E.Wolff@BitWizard.nl That's 5.9*10^12 miles. Your mileage will NOT vary.\__Phone: +31-15-2137555 --(time <-> distance can be converted: lightspeed)-- \____ fax: ..-2138217 We write Linux device drivers for any device you may have! \_______________
- 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/
| |