eSATA Rocks for External Performance

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
Today, I ran my experiment. I have long been impatient with the time it takes to clone a laptop HDD using either a 1394 connection or USB to to an external HDD identical to the internal.

I bit the bullet and got a Vantec eSATA Cardbus PCMCIA device for the laptop, and a Vantec Nex-Star 3 eSATA external case.

I put a 160 GB WD Caviar drive in the Vantec case, and got it all up and running. I then booted with Acronis TrueImage 10's bootable CDR, and proceeded to clone the 160 GB drive in the laptop to the Vantec, and then back again.

Once the external eSATA was formatted and ready to go, the time needed for the cloning operation was 10 minutes.

Before, with the Firewire link to an external 2.5" drive was 55 minutes. That certainly lives up to the generally advertised 5X better performance from eSATA than USB 2.

I ran it again on a second laptop drive (backup) and the same story. 10 minutes for the whole job. This is definitely worthwhile, and for me, signals the eventual demise of external USB 2 drives. I have one, and it is going to go to the "war surpolus" shelf and eventual "trickle down."
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
In the cloning operation it is irrelevant. That is a sole use program run off the bootable CDR - as long as it runs it doesn't matter because nothing else is possible.
 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
Originally posted by: AmigaMan
how's the CPU utilization?
It's basically the same performance and CPU utilization as if you connected to an internal SATA port.
 

jonnyGURU

Moderator <BR> Power Supplies
Moderator
Oct 30, 1999
11,815
104
106
Originally posted by: Accord99
Originally posted by: AmigaMan
how's the CPU utilization?
It's basically the same performance and CPU utilization as if you connected to an internal SATA port.

Yup. eSATA is just a pass through. No bridge chip, nada. It's beautiful.
 

Madwand1

Diamond Member
Jan 23, 2006
3,309
0
76
160 GB / 10m ~ 267 MB/s. That's impossible for a single drive, regardless of price, let alone a laptop drive.

How much data are you actually transferring, and are you sure that it was transferred correctly and that you were doing a fair comparison of data volume / transfer technique with firewire?

You should be constrained by the laptop drive, and firewire, if implemented well, should be close to its performance, on average at least. With your previous test, it's possible that the external laptop drive was the bottleneck.

You should see some improvement with eSATA, but the extent should be much lower than what's reported here.
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
"160 GB / 10m ~ 267 MB/s. That's impossible for a single drive, regardless of price, let alone a laptop drive."

The eSATA is a WDC 7200 RPM 3.5". The laptop drive is a Seagate 5400 RPM, PerRecTech.

"How much data are you actually transferring, and are you sure that it was transferred correctly and that you were doing a fair comparison of data volume / transfer technique with firewire? "

Actually about 40 to 50 MB. The drive is about 25% filled. Transferred correctly? LOL If it were not, I couldn't immediately boot the laptop with the end result. Stem 1 - clone the laptop drive to the eSATA. Step 2 - do it again the other way - colne the eSATA drive to the laptop drive. It booted perfectly and all programs were intact and operable.

"You should be constrained by the laptop drive, and firewire, if implemented well, should be close to its performance, on average at least. With your previous test, it's possible that the external laptop drive was the bottleneck."

F/W is close to its performance. Cloning my desktop drives to like drives via PATA also takes about 10 minutes.

"You should see some improvement with eSATA, but the extent should be much lower than what's reported here. "

eSATA spec is about 1500 - F/W is 400. With the difference of drive speeds, a 5X improvement is not out of line.

Try it and see for yourself.
 

Madwand1

Diamond Member
Jan 23, 2006
3,309
0
76
So you're saying that firewire also does the transfer in about 10 minutes, but eSATA is 5x faster because you read a spec which said that? Which one is it? Is it around the same speed or 5x faster?

Are you comparing just the interface, or the drives? If you're comparing different drives, then how is this a comparison of interface speeds?

I'm not sure what you're saying with this mis-mash of figures and technologies You haven't given a single solid figure. "Around 10 minutes". "Between 40 and 50 MB". MB?...
 

Pariah

Elite Member
Apr 16, 2000
7,357
20
81
160 GB / 10m ~ 267 MB/s. That's impossible for a single drive, regardless of price, let alone a laptop drive.

He said he was imaging a 160GB drive. That doesn't mean the drive was full. When he said 40-50MB, I think he meant GB's. Which would mean about 70MB/s. Still way too fast for any laptop HD, but entering the realm of possibility for highend SCSI drives (which he wasn't using).

eSATA spec is about 1500 - F/W is 400. With the difference of drive speeds, a 5X improvement is not out of line.

You're comparing theoretical (read: unattainable) speeds which don't really mean anything. There's only one conventional HD on the planet (Cheetah 15k.5), that gets anywhere close to the 150MB/s theoretical limit of SATA, and it's a SCSI drive. The fastest SATA drive (Raptor) can only hit about 90MB/s. Your limiting factor is the 5400RPM laptop drive. The fastest on the market peak at around 40MB/s. For E-SATA to be 5x faster, the other interfaces would have to be transferring around 8MB/s. A typical USB 2 enclosure is capable of over 30MB/s, which would be 75% the speed of E-SATA, not 20%. Firewire 400 probably pushes around 35MB/s on a good day.

For desktop drives, E-SATA should be respectably faster (50%+) than firewire and USB2, though no where near 5x faster. For laptop drives, the difference will be quite a bit less (maybe 25%, usually less).
 

Madwand1

Diamond Member
Jan 23, 2006
3,309
0
76
Consumer drives are getting faster. The upcoming Hitachi 1 TB will supposedly hit 100 MB/s. So there's certainly room in the future to take even greater advantage of eSATA.

But to claim that it gives 5x or whatever improvement at present because of an interface spec., is just a misunderstanding that takes you where misleading marketing wants to take you.

Thanks for your post Pariah.
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
I did mean GB - not MB. And the time it takes to clone the drive is also a function of the TrueImage Software. I am merely comparing actual results with actual results. These are real world results - not theoretical.

BTW - the cloning via Firewire to an external identical drive takes 55 minutes. Same actual operation with eSATA takes 10 minutes. Period! That is actual. Same source drive - just different target drives.

I'm not trying to pander to the manufactuirer's specs - merely reporting reality. :)
 

Matthias99

Diamond Member
Oct 7, 2003
8,808
0
0
BTW - the cloning via Firewire to an external identical drive takes 55 minutes. Same actual operation with eSATA takes 10 minutes. Period! That is actual. Same source drive - just different target drives.

Note that TrueImage by default also does compression of your image file. If your FW controller has noticeable CPU usage, it will slow down the whole operation considerably. Try it without compression (or just try copying files within Windows) and see how long it takes.

FW400 has a theoretical max bandwidth of 400Mbps, or 50 megabytes of data per second in sequential transfer. In practice it's often more like 40 due to overhead. USB2 suffers more from overhead and seems to generally be more like 20-25MBps.

A 7200RPM desktop hard drive is going to top out around maybe 70, 75MBps sustained. If eSATA gets you up to 70, I could see a 2x-3x speedup over FW/USB. But it's not going to be five times faster at raw transfers unless your FW/USB controller/enclosure is really really slow for some reason.
 

TBSN

Senior member
Nov 12, 2006
925
0
76
Are there any external videoediting drives that use eSATA yet? Most of the high end ones on the market seem to use Firewire 800....
 
Jan 31, 2002
40,819
2
0
But that's the bonus of eSATA - you don't need to worry about "what controller/chipset/etc" is in your external cage, because there isn't one.

Also, eSATA is just another SATA drive to the computer. Bootable, swappable, formattable ... it's just there.

- M4H
 

TBSN

Senior member
Nov 12, 2006
925
0
76
I'm liking this more and more, and it will factor into my motherboard decision. Essentially, and please excuse the basic questions, eSATA is just an external port into which you can plug an SATA drive, right? Other than that, are there any differences between a normal, internal SATA drive and an external eSATA drive (other than an enclosure and some way to power it I assume...)
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
Originally posted by: Matthias99
Note that TrueImage by default also does compression of your image file.

No, there is no image file when you clone. It is a direct drive to drive, bit by bit copy. I am not "backing up."

To TBSN: eSATA has a slightly different connector - less prone to pull out. There arer bridge cables and links to a mobo SATA port. When that is used, the external drive is seen as an internal drive. The external case has its own power source - that is taken care of by the bridge cable and back panel port.

 

RebateMonger

Elite Member
Dec 24, 2005
11,586
0
0
I've been REALLY happy with external and removable SATA drives so far.

The only "problem" with SATA is that there are issues with REMOVING drives. Some controllers don't show the drives as removable and there's no option to turn off Write Caching or to "Safely Remove" the drive. This could, in theory, cause data loss if the drive is disconnected before all the Writes have been finished.

But I love that the drive controller talks directly to the hard drive, with no conversion chips in between.
 

TBSN

Senior member
Nov 12, 2006
925
0
76
Originally posted by: RebateMonger
I've been REALLY happy with external and removable SATA drives so far.

The only "problem" with SATA is that there are issues with REMOVING drives. Some controllers don't show the drives as removable and there's no option to turn off Write Caching or to "Safely Remove" the drive. This could, in theory, cause data loss if the drive is disconnected before all the Writes have been finished.

Does this mean that SATA (and, more importantly, eSATA) drives are not "hotswappable" (like USB drives are). That was one thing that I excited about concerning this interface...

But I love that the drive controller talks directly to the hard drive, with no conversion chips in between.

I thought that eSATA supported 'hotswapping,' in that a drive can be removed while the OS is still operating. I am very new to a lot of the gritty details of this kind of hardware, so bare with me... SO, in order for a drive to "hotswappable," both the drive and the OS (and the chipset maybe??) must support it, right? So in the above situation, there is a problem with the chipset not supporting hotswapping, or am I totally off base?


[Sorry if my questions don't make any sense, but the only way I know how to figure out how this hardware works is to look it up on wikipedia, browse these excellent forums, and then ask stupid questions...]
 

Pariah

Elite Member
Apr 16, 2000
7,357
20
81
I did mean GB - not MB. And the time it takes to clone the drive is also a function of the TrueImage Software. I am merely comparing actual results with actual results. These are real world results - not theoretical.

Your watch is broken then. The performance you are claiming is simply not possible with the hardware you have. To transfer 40-50GB in 10 minutes as you claim, would take an average transfer rate of 72-90MB/s. There is no 5400RPM laptop on the market that can come remotely close to that performance. There is no debate here, there just isn't. Either you aren't copying as much data as you are claiming, or it is taking longer than you think it is.

A 7200RPM desktop hard drive is going to top out around maybe 70, 75MBps sustained. If eSATA gets you up to 70, I could see a 2x-3x speedup over FW/USB. But it's not going to be five times faster at raw transfers unless your FW/USB controller/enclosure is really really slow for some reason.

One of the drives he is using in the cloning process is a 5400RPM laptop drive. The fastest 2.5" 5400RPM drives I have seen peak at about 40MB's.
 

Minerva

Platinum Member
Nov 18, 1999
2,127
19
81
Is this a partition containing a paging file? I've seen this happen where the data set was n+p where p=paging file. So when the % complete reaches where the paging file is it immediately springs to 100% complete! This makes the data rate seem much faster than it is.

Problem with ESATA is hot pluggability. Most SATA ports don't support hot plugging both ways.
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
Originally posted by: Minerva
Is this a partition containing a paging file? I've seen this happen where the data set was n+p where p=paging file. So when the % complete reaches where the paging file is it immediately springs to 100% complete! This makes the data rate seem much faster than it is. Problem with ESATA is hot pluggability. Most SATA ports don't support hot plugging both ways.

The drive does contain a small partition with the page file. It copies the page file verbatim - otherwise the target drive would not have it identical to the source drive.

Hot plugability depends on where it is connected. In this case, it is to a specific eSATA port in a Vantec Cardbus PCMCIA device. It shows in D/M Properties as "Optimized for Quick Removal" with Write Caching off. If it were connected to a morherboard SATA port via a bridge cable to external port, it would be seen as an internal drive "Optimized for Performance with Write Caching turned on. The "Quick Removal" option would be grayed out and not available.

The drive in the laptop is a Seagate 160 GB, 5400 RPM, Perpendicular Recording - and they are empirically, a bit faster than regular 5400 drives. I have had both. The laptop started with 60GB, then 100, then 120, then 160. All 5400. The 160 has empirically been the fastest.

As for time, I am using the time stated by TrueImage as it starts the clone job.

What I might do is run the same operation with the same drives and use the USB connection instead. It is a "either/or" combo case.



 

Minerva

Platinum Member
Nov 18, 1999
2,127
19
81
I don't know if a disk imaging program is a good benchmark. Perhaps if it's real world performance it is.

With Symantec Ghost and high compression I went from mid 5500's to low 7200's (in MB/min) just by increasing the cpu (QX6700) core from 3.2 to 3.4 GHz. Ghost is a DOS program so I know it's not using all four cores but that's a nice jump!

A clone SHOULD copy everything bit for bit. I haven't tried it with a 16GB paging file in xp64 with clone but image with compression it shot over as the paging file is zero data.
 

corkyg

Elite Member | Peripherals
Super Moderator
Mar 4, 2000
27,370
239
106
Originally posted by: Minerva
I don't know if a disk imaging program is a good benchmark. Perhaps if it's real world performance it is.

Well, as a long time user with many daily requirements, benchmarks are useful mainly for shoppers. I believe in dealing with empirical results where the work is done. The reason I bought the eSATA setup was not for disk imaging - but for drive cloning. Images have to be restored. Cloned bootable drives are immediately bootable w/o restoration.



 

Minerva

Platinum Member
Nov 18, 1999
2,127
19
81
We use benchmarks really to determine if the hardware is functioning properly. If used correctly this is a valuable tool. Used incorrectly (like manufacturers tuning their firmware to falsely inflate winbench scores) is quite counterproductive.

When a particular number seems to exceed the physical capabilities of the medium investigation into why this happens is in order as there are no miracles. :)

With firewire and USB there's a lot of command overhead with translation amongst interfaces. This can slow down certain processes enormously. Within the windows environment the windows cache itself helps considerably but this will not be available outside of the windows environment.

Finally, yes I know the difference between a clone and an image. We image for recovery purposes. Images are kept on a server that's available from our BOOTP tables so recovery is a snap with the lan rom selection. :)
 

Slickone

Diamond Member
Dec 31, 1999
6,120
0
0
If I use an enclosure that accepts SATA or PATA interface internally, but is eSATA out, if using a PATA drive I would be running slower than if using a SATA drive, correct?