[mythtv-users] SATA Drive Failure

Allan Stirling Dibblahmythml0015 at pendor.org
Sat Jun 25 11:06:14 UTC 2005


Kelly wrote:
> Think I've found a source of my constant crashing. I have a Maxtor
> SATA 250GB drive. I started poking around google and came up with the
> hdparm test. I ran it and got
> 
> Timing cached reads:   816 MB in  2.01 seconds 
> 
> A respectable IDE drive and do 3000mb in 2.00 seconds. 
> 
> So my drive is running at 20% efficiency. That will make a difference. 
> 
Ummm... No. You're reading the results wrongly, I'm afraid :)
I would seriously doubt you'd see 1.5Gb / sec outside the datacentre.
150Mb / sec... _maybe_.

Cached reads are just that - It reads the same sector over and over. 
This means it's already in drive's read cache. So what you're seeing 
here is the transfer rate from the controller without (much) SATA 
latency, no seek latency, no rotational latency. That's not the test you 
want. You want 'buffered disk reads'. This is the (almost real) raw 
performance of the drive.

Here's the results from my array, which is a 8 disk RAID5, spread over 2 
SATA controllers on a PCI-X dual Xeon (yes, I believe overkill _is_ just 
a word):

hdparm -Tt /dev/sda

/dev/sda:
  Timing cached reads:   2204 MB in  2.00 seconds = 1101.07 MB/sec
  Timing buffered disk reads:  374 MB in  3.01 seconds = 124.31 MB/sec

The 'buffered disk reads' is the line you're looking for. And 124Mb / 
sec is quite good for this type of setup.

Cheers,

Allan.


More information about the mythtv-users mailing list