RE: [flalug] backup drive

From: Michael Worsham (michael.worsham@mci.com)
Date: Tue Dec 09 2003 - 15:41:04 EST


Anytime a file is altered, moved, copied, (locally or externally via NFS,
tape, etc), it will utilize CPU power... it just depends on the amount of
data you are trying to pass. This is even true w/ SCSI and SATA drives as
well. NAS layouts are usually less cpu bound, but there is still some
residual cpu usage anyway.

-- M

-----Original Message-----
From: flalug@nks.net [mailto:flalug@nks.net]On Behalf Of Eben King
Sent: Tuesday, December 09, 2003 2:47 PM
To: flalug@nks.net
Subject: Re: [flalug] backup drive

On Tue, 9 Dec 2003, perrin wrote:

> > Oh, I thought that by "image" you meant like Ghost makes under Windows,
or
> > like "cat /dev/hda > file" would make. You meant duplicate the drive.
> > Right, that'll work. That's basically what I'm doing now.
>
> Last I checked, Ghost makes an exact duplicate that doesn't have any
> additional overhead on the destination media.

Surely it puts some signature bytes at the beginning or end. Otherwise it
just has the filename and your word to go on,

> I suppose I could be mistaken.

Well, if you have Ghost (I don't), you could test it...

> > Something must be wrong, though -- CPU use rises to ~50% during the
copy.
> > Both drives use DMA, according to hdparm. I didn't think CPU was
supposed
> > to be used during a DMA transfer.
>
> Copying uses processing power because the processor is making decisions to
> manage reading and writing. DMA only allows other devices to directly
> access data from the drives without having to go through the processor.
> Something still needs to control the drives, and if there aren't any
> additional devices doing so that task falls to the CPU.

What kind of devices would control an IDE hard drive? Would a SCSI hard
drive exhibit the same phenomenon?

--
-eben    ebQenW1@EtaRmpTabYayU.rIr.OcoPm    home.tampabay.rr.com/hactar

A: Because it looks dumb and is hard to read. Q: Why is top-posting wrong? -- from lots42@xxx.com



This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 19:58:58 EDT