Serial ATA - So what?

Serial ATA - So what?

Post by George Bel » Wed, 07 May 2003 07:38:40



Could somebody please explain to me what is so great about the new
Serial ATA.  What good is a faster interface to the harddrive if the
present one(ATAPI 133 the latest?) is already far faster than harddrives
can access the data? My impression was that the bottleneck is in how
fast the mechanical heads can move and how fast the disks can rotate,
and that this was far slower than the maximum speeds on the bus which
connects the harddrive to the rest of the system.

Thanks,
George

 
 
 

Serial ATA - So what?

Post by M?ns Rullg? » Wed, 07 May 2003 08:17:20



> Could somebody please explain to me what is so great about the new
> Serial ATA.  What good is a faster interface to the harddrive if the
> present one(ATAPI 133 the latest?) is already far faster than
> harddrives can access the data? My impression was that the bottleneck
> is in how fast the mechanical heads can move and how fast the disks
> can rotate, and that this was far slower than the maximum speeds on
> the bus which connects the harddrive to the rest of the system.

That's all true.  There are still advantages.

- Disks will likely become faster in the future, and SATA is ready for
this.

- The SATA signaling is less sensitive to interference than parallel
ATA.  My machine is apparently rather noisy, so with old ATA, I can
only use ATA-33, but SATA works perfectly at the faster speeds.

- The cables are *much* easier to work with.  Since they are thinner,
they do not obstruct air flow very much.  The smaller connectors also
make it possible to have more ports on a single board, so there is no
need to use master/slave setups.  This is good for performance.

Did I miss anything important?

--
M?ns Rullg?rd


 
 
 

Serial ATA - So what?

Post by Ma.. » Wed, 07 May 2003 09:18:00




> > Could somebody please explain to me what is so great about the new
> > Serial ATA.  What good is a faster interface to the harddrive if the
> > present one(ATAPI 133 the latest?) is already far faster than
> > harddrives can access the data? My impression was that the bottleneck
> > is in how fast the mechanical heads can move and how fast the disks
> > can rotate, and that this was far slower than the maximum speeds on
> > the bus which connects the harddrive to the rest of the system.

> That's all true.  There are still advantages.

> - Disks will likely become faster in the future, and SATA is ready for
> this.

> - The SATA signaling is less sensitive to interference than parallel
> ATA.  My machine is apparently rather noisy, so with old ATA, I can
> only use ATA-33, but SATA works perfectly at the faster speeds.

> - The cables are *much* easier to work with.  Since they are thinner,
> they do not obstruct air flow very much.  The smaller connectors also
> make it possible to have more ports on a single board, so there is no
> need to use master/slave setups.  This is good for performance.

> Did I miss anything important?

> --
> M?ns Rullg?rd


I am hoping that SATA has a higher-level interface, like ATAPI (or SCSI).
Does anyone know if that is true?  If so, I expect SATA drives should
operate more reliably (under DMA).  Also the thin cables will make
it easier to make huge RAID arrays in tight spaces, like rackmount cases.

  Richard

 
 
 

Serial ATA - So what?

Post by Whoeve » Wed, 07 May 2003 09:20:24




> > Could somebody please explain to me what is so great about the new
> > Serial ATA.  What good is a faster interface to the harddrive if the
> > present one(ATAPI 133 the latest?) is already far faster than
> > harddrives can access the data? My impression was that the bottleneck
> > is in how fast the mechanical heads can move and how fast the disks
> > can rotate, and that this was far slower than the maximum speeds on
> > the bus which connects the harddrive to the rest of the system.

> That's all true.  There are still advantages.

> - Disks will likely become faster in the future, and SATA is ready for
> this.

> - The SATA signaling is less sensitive to interference than parallel
> ATA.  My machine is apparently rather noisy, so with old ATA, I can
> only use ATA-33, but SATA works perfectly at the faster speeds.

> - The cables are *much* easier to work with.  Since they are thinner,
> they do not obstruct air flow very much.  The smaller connectors also
> make it possible to have more ports on a single board, so there is no
> need to use master/slave setups.  This is good for performance.

> Did I miss anything important?

The implications of having a cache on the disk?

- Show quoted text -

 
 
 

Serial ATA - So what?

Post by Eben Kin » Wed, 07 May 2003 09:36:53





> > > Could somebody please explain to me what is so great about the new
> > > Serial ATA.

> > - Disks will likely become faster in the future, and SATA is ready for
> > this.

> > - The SATA signaling is less sensitive to interference than parallel
> > ATA.  My machine is apparently rather noisy, so with old ATA, I can
> > only use ATA-33, but SATA works perfectly at the faster speeds.

> > - The cables are *much* easier to work with.  Since they are thinner,
> > they do not obstruct air flow very much.  The smaller connectors also
> > make it possible to have more ports on a single board, so there is no
> > need to use master/slave setups.  This is good for performance.

> > Did I miss anything important?

> The implications of having a cache on the disk?

Is that cache significantly different than the 2MB-8MB+ caches currently on
ATAPI disks?
 
 
 

Serial ATA - So what?

Post by B. Joshua Rose » Wed, 07 May 2003 10:18:42




>> Could somebody please explain to me what is so great about the new
>> Serial ATA.  What good is a faster interface to the harddrive if the
>> present one(ATAPI 133 the latest?) is already far faster than
>> harddrives can access the data? My impression was that the bottleneck
>> is in how fast the mechanical heads can move and how fast the disks can
>> rotate, and that this was far slower than the maximum speeds on the bus
>> which connects the harddrive to the rest of the system.

> That's all true.  There are still advantages.

> - Disks will likely become faster in the future, and SATA is ready for
> this.

> - The SATA signaling is less sensitive to interference than parallel
> ATA.  My machine is apparently rather noisy, so with old ATA, I can only
> use ATA-33, but SATA works perfectly at the faster speeds.

> - The cables are *much* easier to work with.  Since they are thinner,
> they do not obstruct air flow very much.  The smaller connectors also
> make it possible to have more ports on a single board, so there is no
> need to use master/slave setups.  This is good for performance.

> Did I miss anything important?

There are a couple of important things you missed, the cables can be 1
meter long and SATA supports hot swap. The next version, SATA II will add
some SCSI like capbilites also which will further improve performance.
 
 
 

Serial ATA - So what?

Post by Stace » Wed, 07 May 2003 10:32:45



> Could somebody please explain to me what is so great about the new
> Serial ATA.  What good is a faster interface to the harddrive if the
> present one(ATAPI 133 the latest?) is already far faster than harddrives
> can access the data? My impression was that the bottleneck is in how
> fast the mechanical heads can move and how fast the disks can rotate,
> and that this was far slower than the maximum speeds on the bus which
> connects the harddrive to the rest of the system.

Smaller cables is all for now. Even ata133 over ata100 is a joke in real
use.

--

  Stacey

 
 
 

Serial ATA - So what?

Post by Christopher Brown » Wed, 07 May 2003 10:52:42



Quote:> Could somebody please explain to me what is so great about the new
> Serial ATA.  What good is a faster interface to the harddrive if the
> present one(ATAPI 133 the latest?) is already far faster than
> harddrives can access the data? My impression was that the bottleneck
> is in how fast the mechanical heads can move and how fast the disks
> can rotate, and that this was far slower than the maximum speeds on
> the bus which connects the harddrive to the rest of the system.

There are two notable wins in it:

1.  The scheme offers a SCSI-like "tagged queueing" approach that
    should let it become more efficient than IDE has traditionally
    been;

2.  SATA provides a more convenient cabling system supporting longer
    cables, and a _much_ nicer form factor, thus making "RAID-like"
    usage easier to manage.

It's a nice win if you can readily hook up 6 drives without needing to
warp your cables...
--
If this was helpful, <http://svcs.affero.net/rm.php?r=cbbrowne> rate me
http://www.ntlug.org/~cbbrowne/hardware.html
Rules of  the Evil Overlord  #123. "If I  decide to hold a  contest of
skill  open to  the general  public, contestants  will be  required to
remove their  hooded cloaks and  shave their beards  before entering."
<http://www.eviloverlord.com/>

 
 
 

Serial ATA - So what?

Post by Christopher Brown » Wed, 07 May 2003 10:52:43



Quote:> The implications of having a cache on the disk?

You mean on the disk controller?  Much like the implications of the
fact that even IDE disks /already/ have caches on their controllers?
--

http://cbbrowne.com/info/wp.html
"Other than that, Mrs. Lincoln, how was the play?"
 
 
 

Serial ATA - So what?

Post by Whoeve » Wed, 07 May 2003 14:51:24







> > > > Could somebody please explain to me what is so great about the new
> > > > Serial ATA.

> > > - Disks will likely become faster in the future, and SATA is ready for
> > > this.

> > > - The SATA signaling is less sensitive to interference than parallel
> > > ATA.  My machine is apparently rather noisy, so with old ATA, I can
> > > only use ATA-33, but SATA works perfectly at the faster speeds.

> > > - The cables are *much* easier to work with.  Since they are thinner,
> > > they do not obstruct air flow very much.  The smaller connectors also
> > > make it possible to have more ports on a single board, so there is no
> > > need to use master/slave setups.  This is good for performance.

> > > Did I miss anything important?

> > The implications of having a cache on the disk?

> Is that cache significantly different than the 2MB-8MB+ caches currently on
> ATAPI disks?

Well, caches are getting bigger and will continue to do so, but: the OP
asked why one needs fast interfaces to disks when the disk itself is so
slow to take data off the platters. My point is that this does not
necessarily limit the data rate that the disk drive can achieve because
the cache can provide much faster data rates for small bursts of data
to/from the cache.

- Show quoted text -

 
 
 

Serial ATA - So what?

Post by Eben Kin » Wed, 07 May 2003 15:44:52








> > > > > Could somebody please explain to me what is so great about the new
> > > > > Serial ATA.

> > > > - Disks will likely become faster in the future, and SATA is ready
for
> > > > this.

> > > > - The SATA signaling is less sensitive to interference than parallel
> > > > ATA.  My machine is apparently rather noisy, so with old ATA, I can
> > > > only use ATA-33, but SATA works perfectly at the faster speeds.

> > > > - The cables are *much* easier to work with.  Since they are
thinner,
> > > > they do not obstruct air flow very much.  The smaller connectors
also
> > > > make it possible to have more ports on a single board, so there is
no
> > > > need to use master/slave setups.  This is good for performance.

> > > > Did I miss anything important?

> > > The implications of having a cache on the disk?

> > Is that cache significantly different than the 2MB-8MB+ caches currently
on
> > ATAPI disks?

> Well, caches are getting bigger and will continue to do so, but: the OP
> asked why one needs fast interfaces to disks when the disk itself is so
> slow to take data off the platters. My point is that this does not
> necessarily limit the data rate that the disk drive can achieve because
> the cache can provide much faster data rates for small bursts of data
> to/from the cache.

Yes, caches are a Good Thing.  But how are serial ATA disks' cache
implentations different from ATA disks'?  I believe the OP's point was "what
makes serial ATA better than ATA, considering that ATA's speed isn't the
bottleneck?".  ICBW, of course...
 
 
 

Serial ATA - So what?

Post by M?ns Rullg? » Wed, 07 May 2003 16:55:44



Quote:> >> Could somebody please explain to me what is so great about the new
> >> Serial ATA.  What good is a faster interface to the harddrive if the
> >> present one(ATAPI 133 the latest?) is already far faster than
> >> harddrives can access the data? My impression was that the bottleneck
> >> is in how fast the mechanical heads can move and how fast the disks can
> >> rotate, and that this was far slower than the maximum speeds on the bus
> >> which connects the harddrive to the rest of the system.

> > That's all true.  There are still advantages.

> > - Disks will likely become faster in the future, and SATA is ready for
> > this.

> > - The SATA signaling is less sensitive to interference than parallel
> > ATA.  My machine is apparently rather noisy, so with old ATA, I can only
> > use ATA-33, but SATA works perfectly at the faster speeds.

> > - The cables are *much* easier to work with.  Since they are thinner,
> > they do not obstruct air flow very much.  The smaller connectors also
> > make it possible to have more ports on a single board, so there is no
> > need to use master/slave setups.  This is good for performance.

> > Did I miss anything important?

> There are a couple of important things you missed, the cables can be 1
> meter long and SATA supports hot swap. The next version, SATA II will add
> some SCSI like capbilites also which will further improve performance.

That's right, but does Linux support hot-swapping SATA drives?  Do
even the controllers support it?

--
M?ns Rullg?rd

 
 
 

Serial ATA - So what?

Post by M?ns Rullg? » Wed, 07 May 2003 16:54:26



> > > > > Could somebody please explain to me what is so great about the new
> > > > > Serial ATA.

> > > > - Disks will likely become faster in the future, and SATA is ready for
> > > > this.

> > > > - The SATA signaling is less sensitive to interference than parallel
> > > > ATA.  My machine is apparently rather noisy, so with old ATA, I can
> > > > only use ATA-33, but SATA works perfectly at the faster speeds.

> > > > - The cables are *much* easier to work with.  Since they are thinner,
> > > > they do not obstruct air flow very much.  The smaller connectors also
> > > > make it possible to have more ports on a single board, so there is no
> > > > need to use master/slave setups.  This is good for performance.

> > > > Did I miss anything important?

> > > The implications of having a cache on the disk?

> > Is that cache significantly different than the 2MB-8MB+ caches currently on
> > ATAPI disks?

> Well, caches are getting bigger and will continue to do so, but: the OP
> asked why one needs fast interfaces to disks when the disk itself is so
> slow to take data off the platters. My point is that this does not
> necessarily limit the data rate that the disk drive can achieve because
> the cache can provide much faster data rates for small bursts of data
> to/from the cache.

What difference does 8 MB cache on the disk make, when I anyway have
400 MB cached in RAM?

--
M?ns Rullg?rd

 
 
 

Serial ATA - So what?

Post by Juha Laih » Thu, 08 May 2003 02:42:00



Quote:>Could somebody please explain to me what is so great about the new
>Serial ATA.  What good is a faster interface to the harddrive if the
>present one(ATAPI 133 the latest?) is already far faster than harddrives
>can access the data? My impression was that the bottleneck is in how
>fast the mechanical heads can move and how fast the disks can rotate,
>and that this was far slower than the maximum speeds on the bus which
>connects the harddrive to the rest of the system.

Well, ATA-133 is ok - for one disk. How about creating an array of five
disks on the same channel? I'd guess you're hitting a bus bottleneck
quite soon. Ok, you don't see five disks in regular desktops - but it
seems SATA has also other features that are intended to serve in systems
other than regular desktops.
--
Wolf  a.k.a.  Juha Laiho     Espoo, Finland

         PS(+) PE Y+ PGP(+) t- 5 !X R !tv b+ !DI D G e+ h---- r+++ y++++
"...cancel my subscription to the resurrection!" (Jim Morrison)
 
 
 

Serial ATA - So what?

Post by Eben Kin » Thu, 08 May 2003 03:16:34



Quote:

> > > > > > Could somebody please explain to me what is so great about the
new
> > > > > > Serial ATA.

> > > > The implications of having a cache on the disk?

> What difference does 8 MB cache on the disk make, when I anyway have
> 400 MB cached in RAM?

Does the 8MB have lower latency?  I'm prettty sure it does not have higher
throuput...

Does cache on the disk make no difference, then?  Would a disk without cache
perform as well as one with cache?

 
 
 

1. Serial ATA and IDE problems with 2.4.2x

Hello world,

I'm trying to get an AMD64 system to work with linux.  I've got both
Mandrake and Suse distro's.  Mandrake is version 9.2 and uses the 2.4.24
kernel, Suse is version 9.0 and uses the 2.4.21 kernel.

My motherboard has 2 EIDE ports and 2 SATA connectors.  I started with
one HD on the EIDE (Maxtor 40G) and one HD on the SATA (Segate, 80G).
With the bios set up to see the SATA as the first drive, Mandrake would
load and boot on the SATA (but it won't work with the modem, so I went
and got Suse).

Suse wouldn't load onto the SATA, and when I got the Patch-CD it would
hang while loading the via_sata module (maybe that's sata_via, I
forget).  So then I got another Maxtor drive and plugged it in as a
slave behind the first one.

At that point, with 3 drives connected, neither Suse nor Mandrake could
get past the sata module - the boot process just hangs.  So I removed
the SATA drive, and loaded Suse onto the slave Maxtor.  That worked.
And I bet Mandrake would work too - the problem is in the kernel.

I'd like to get the serial ATA drive working, but it looks like I'll
have to patch some code.  The hang point is in the scsi driver, which
doesn't make sense to me, but I guess I don't know how the kernel hangs
together.  Who should I ask for help and what doc's do I need to read
to get up to speed on fixing this problem?

Any help appreciated!

Patience, persistence, truth,
Dr. mike

2. Analog (log file analyzer) question

3. Serial ATA

4. recompiling a program while it is running

5. Adaptec SERIAL ATA RAID 2410SA on RedHat Linux

6. bash, simplify "x=$(basename $PWD); y=${x#*-}"?

7. FW: LSI Megaraid Serial ATA driver ...

8. rm'ing files such as -C

9. serial ATA support

10. Linux Serial ATA Driver, late Q3 Release Schedule

11. Serial ATA

12. serial ATA (SATA) GNU/Linux support?

13. Solaris 9 and serial ATA