building a RAID system - yup
xyzzy at speakeasy.org
Thu Oct 9 14:40:02 EDT 2003
On Thu, 9 Oct 2003, Robert G. Brown wrote:
> It probably refers to burst delivery out of its 8 MB cache. The actual
> sustained bps speed is a pure matter of N*2*\pi*\R*f/S, where N = number
A hard drive only reads from one head at a time. It's not possible to align
every head with each other to such a degree that every track in a cylinder is
readable at once. If you look at a given drive family of drives, each
different sized drive is the same basic hardware with more discs/heads. For
instance Seagate's Cheetah 15K.3 family
(http://www.seagate.com/docs/pdf/datasheet/disc/ds_cheetah15k.3.pdf) has the
exact same internal transfer rate (609-891 megabits/sec) for the 18 GB model
with 2 heads, the 36GB with 4 heads, and the 73GB with 8.
> read radius, and S is the linear length per bit. This is an upper
> bound. Similarly average latency (seek time) is something like 1/2f,
> the time the platter requires to move half a rotation.
The average latency is indeed 1/2 the rotational period. For a 7200 RPM drive
it is 4.16 ms, for a 15k RPM drive it's 2 ms. Seek time is something
completely different. It's how long it takes the head to move from one track
to another. It does not included the latency. You might see track-to-track,
full stroke, and average seek times in a datasheet.
> I should also point out that since we've been using the RAIDs we have
> experienced multidisk failures that required restoring from backup on
> more than one occasion. The book value probability for even one
I've had one multidisk failure in a RAID5 system. It was after moving into a
new building, one array had three out of six disks fail to spin up. Of course
I had anticipated this, and made a backup, to tape, just before the move.
None of the tapes were damaged in transit. I've had several single drive
failures. I've never seen anyone with significant number of drive-years of
experience say they've never seen a drive fail. And no manufacture has a
failure rate anywhere near 0%.
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
More information about the Beowulf