"Charles T. Smith" <[email protected]> writes:

Hi, a quick question for the experts.

I've gathered the impression that the interface technology (usb, sata,
pata) for disk drives is really irrelevant - for a single drive - because disks can't deliver data fast enough to run into an interface bottleneck.

USB2 is slow enough that it’s a serious bottleneck.

Maybe that assumption's not correct... anyway, I see an ad for a Seagate external 2.5" drive which uses "USB 3.0 Super-speed, up to 10x faster
than USB 2.0". While that's likely true about USB 3.0, isn't it
cynically, unethically misleading about the effect on the drive's performance? Or, why would Seagate supply drives with USB 3.0 technology?

USB3-connected drives are *much* faster than USB2, based on the drives I
have.

--
http://www.greenend.org.uk/rjk/

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Thanks, gentlemen.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 8 Aug 2015 07:06:11 +0000 (UTC), "Charles T. Smith" <[email protected]> wrote:

Hi, a quick question for the experts.

I've gathered the impression that the interface technology (usb, sata,
pata) for disk drives is really irrelevant - for a single drive - because >disks can't deliver data fast enough to run into an interface bottleneck.

Maybe that assumption's not correct... anyway, I see an ad for a Seagate >external 2.5" drive which uses "USB 3.0 Super-speed, up to 10x faster
than USB 2.0". While that's likely true about USB 3.0, isn't it
cynically, unethically misleading about the effect on the drive's >performance? Or, why would Seagate supply drives with USB 3.0 technology?

I once did the numbers regarding cylinder, spin-rate, etc, but they say
that these days those parameters are irrelevant ...

They're not irrelevant. Modern desktop spinning disk can hit transfer
rates from the platter well over 100MB/s, at least for sequ3ential
reads. Enterprise drives are often in the 300MB/s range. USB2.0 is
limited to a theoretical 60MB/s, but as in practical terms rarely
achieves more than about half that. Also I/O to the cache onboard the
drives can happen faster than that. So USB 2.0 really is a bottle
neck when performing bulk (sequential) I/O to a disk drive. For
random I/O, it makes little difference.

Now as to whether or not USB 3.0 is *really* ten times faster is a
different question (sure, the interface is ~10 times faster), but
there's no doubt that things like backups to an external drive can run considerably faster with a USB 3.0 connection in many cases. Backups
are a major use case for external drives, so sequential performance
*is* important.

OTOH, the practical difference between SATA 1/2/3 and USB 3.0 is
fairly minor. OTTH, the difference between SATA 1 and SATA 3 is only
a factor of four, and SATA 1 is 150MB/s anyway, far faster than USB
2.0 (and SATA tends to use more of its theoretical bandwidth).
Likewise the fastest PATA (133MB/s) interfaces have not been too far
off keeping up with desktop drives until recently.

In a sense the practical difference between PATA-133, SATA-1/2/3 and
USB 3.0 are modest, but USB 2.0 (or say PATA-33) is actually
substantially slower.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 08 Aug 2015 03:01:07 -0500, Robert Wessel
<[email protected]> wrote:

On Sat, 8 Aug 2015 07:06:11 +0000 (UTC), "Charles T. Smith" ><[email protected]> wrote:

Hi, a quick question for the experts.

I've gathered the impression that the interface technology (usb, sata, >>pata) for disk drives is really irrelevant - for a single drive - because >>disks can't deliver data fast enough to run into an interface bottleneck.

Maybe that assumption's not correct... anyway, I see an ad for a Seagate >>external 2.5" drive which uses "USB 3.0 Super-speed, up to 10x faster
than USB 2.0". While that's likely true about USB 3.0, isn't it
cynically, unethically misleading about the effect on the drive's >>performance? Or, why would Seagate supply drives with USB 3.0 technology?

I once did the numbers regarding cylinder, spin-rate, etc, but they say >>that these days those parameters are irrelevant ...

They're not irrelevant. Modern desktop spinning disk can hit transfer
rates from the platter well over 100MB/s, at least for sequ3ential
reads. Enterprise drives are often in the 300MB/s range. USB2.0 is
limited to a theoretical 60MB/s, but as in practical terms rarely
achieves more than about half that. Also I/O to the cache onboard the
drives can happen faster than that. So USB 2.0 really is a bottle
neck when performing bulk (sequential) I/O to a disk drive. For
random I/O, it makes little difference.

Now as to whether or not USB 3.0 is *really* ten times faster is a
different question (sure, the interface is ~10 times faster), but
there's no doubt that things like backups to an external drive can run >considerably faster with a USB 3.0 connection in many cases. Backups
are a major use case for external drives, so sequential performance
*is* important.

OTOH, the practical difference between SATA 1/2/3 and USB 3.0 is
fairly minor. OTTH, the difference between SATA 1 and SATA 3 is only
a factor of four, and SATA 1 is 150MB/s anyway, far faster than USB
2.0 (and SATA tends to use more of its theoretical bandwidth).
Likewise the fastest PATA (133MB/s) interfaces have not been too far
off keeping up with desktop drives until recently.

In a sense the practical difference between PATA-133, SATA-1/2/3 and
USB 3.0 are modest, but USB 2.0 (or say PATA-33) is actually
substantially slower.

And here's an on-point Macworld article from a few years ago:

http://www.macworld.com/article/2039427/how-fast-is-usb-3-0-really-.html

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)