|
|||||||||
|
Thread Tools | Search this Thread |
May 22nd, 2009, 05:22 PM | #1 |
Inner Circle
Join Date: Jan 2006
Posts: 2,290
|
Benefits of external e-sata drive
Is there a real world avantage to using and external hard drive connected via e-sata cable to a 54mm express card on an older 1.66 ghz dual core laptop? Or am I just making work for myself? The bus speed is like 533mhz. The external hdd would be for video files.
|
May 23rd, 2009, 07:23 AM | #2 |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
In comparison to what?
Internal SATA disk: No. External Firewire disk: Yes. FW800 is slower, FW400 is far slower. External USB disk: YES. An enormous difference. |
May 23rd, 2009, 07:49 AM | #3 |
Kino-Eye
Join Date: Jul 2005
Location: Boston, MA
Posts: 457
|
A while back I did my own benchmark testing comparing external SATA, FireWire 800, FireWire 400, and internal SATA using my MacBook Pro, and here are the results:
Comparing MacBook Pro external drive performance The test confirms for me that unless you’re using a hardware RAID, the fastest single disk performance you’re going to see with a MacBook Pro is with an external 7200RPM SATA drive connected via a SATA interface card. And I did not compare to USB 2.0, however, it would have been even slower, in spite of the specifications on paper, FireWire is superior to USB 2.0 for hard drive transfer rates. An interesting thing I'd like to compare in the future would be a hardware RAID-0 via FireWire 800 vs. a single external SATA drive via SATA interface. Specs on paper tell you nothing, you have to test and see performance under real world conditions.
__________________
David Tames { blog: http://Kino-Eye.com twitter: @cinemakinoeye } |
May 23rd, 2009, 09:40 AM | #4 |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
David, you made a good point about RPM's of disks and especially in notebooks. I don't have figures readily available for 4200 or 5400 disks, since I consider them unsuitable for editing and have not tried to benchmark them.
Brian, my earlier response to you only applies to 7200 RPM disks. A rough rule of thumb is that the various kinds of disks can achieve the following sustained transfer rates (depending on fill rate and a number of other factors). This applies to 7200 RPM disks only. SATA/eSATA: 80 - 100 MB/s Firewire 800 : 45 - 55 MB/s Firewire 400 : 25 - 35 MB/s USB 2.0 : 18 - 25 MB/s (depending on the number of other USB devices attached) For capturing all disks are fast enough. For editing even a single SATA or eSATA disk can be overwhelmed during editing. For comparison, a two disk SATA raid0 achieves around 160-180 MB/s, a 10 disk or larger SATA raid3 array can achieve 800 MB/s easily with the correct controller. |
May 23rd, 2009, 01:27 PM | #5 |
Inner Circle
Join Date: Jan 2006
Posts: 2,290
|
So a card like this for my notebook port OWC ExpressCard/34 eSATA SATA I/II Add-O... (EXP34SATAIIP1) at OWC
an eSata connector cable, an external case with eSata port, and of course a hard drive, and I'm good to go? There's surprising little info on this topic on the web. Very odd since it seems like such a great solution. Also, my slower bus speed won't be the limiting factor? It's only 667mhz |
May 23rd, 2009, 02:09 PM | #6 |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
Correct. The speed of your FSB 667 MHz is not a limiting factor and even irrelevant.
|
May 24th, 2009, 08:00 AM | #7 |
Major Player
Join Date: Mar 2005
Location: Ontario, Canada
Posts: 916
|
Harm, we've been researching NAS and gigabit network performance for nearly a year now which included extensive SATA testing in RAID 0 and 5 using various cards and onboard mobo chipsets. These tests are PC based but have used SMB1 and SMB2 on both 32bit and 64 bit platforms. The most we've ever seen sustained from a single drive sustained on large file transfers is in the 60MB/s area which is consistent with practical theoritical predictions. Our testing is based on sample files transfers as well as encodes which are timed and calculated so based on average results. We don't typically test with empty drives so the platter peripheral speeds are not overly skewing our results. Our common test file set is an 5GB SxS card dump/file set with files up to 2GB in size.
We are now seeing speeds over gigabit LAN that average 103MB/s read and 93MB/s write to a gigabit LAN connected NAS (RAID5, 5TB) unit which we've found is a very effective collaborative editing setup. To sustain these speeds requires a minimum of RAID0 x 3 drives on our editing workstations using SMB2 (Vista 64 bit). Cheers, Dennis Wood www.cinevate.com |
May 24th, 2009, 08:26 AM | #8 |
I routinely run RAID0 on an Addonics SATA/RAID PCIe (eSATA) card in my laptop. Sequential read/write speeds test at 96.7/90.96 with Diskmark on a RAID0x2. By comparison, the onboard single SATA 7200RPM Hitachi hard drive tests at 36.3/27.8.
|
|
May 24th, 2009, 11:59 AM | #9 |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
Dennis, those are very good figures for a NAS. If you start teaming NIC's you can even further improve on those. I only use the NAS (7 x 1.5 TB) for backup/storage, not for video editing, but I have 16 TB internally, of which 12 TB in a raid30, for media and projects, so the need for me to use a NAS for NLE work is not urgent.
Nevertheless, your results are very positive and look like a very good alternative to internal storage. |
June 2nd, 2009, 10:26 PM | #10 |
Inner Circle
Join Date: Jan 2007
Location: Woodinville, WA USA
Posts: 3,467
|
Hey Harm (and all you other SATA experts out there) --
I just picked a couple of 2TB eSATA drives to use for backup and archive, and I ordered a 4-port eSATA card and cables to go with them. But on my Areca 1231, I've got 5 unused internal SATA connectors inside the PC case. Would I get better performance by simply running these connectors outside and a SATA to eSATA cable straight into the drives, rather than using the PCI-X eSATA card? I guess the advantage would be having the Areca manage all of the drives (the external drives wouldn't be part of the internal RAID array, as I picture it). Would there be any difference in speed/reliability? Any reasons to do or not do either option? |
June 3rd, 2009, 04:18 AM | #11 |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
Hi Adam,
I do not think you will notice a large difference in performance and since it is for backup purposes it may not be very relevant, but IIRC the 1231 uses an IOP 341, which is a pretty good chip. The 4 port card you ordered may not achieve the same performance, especially if you have more cache installed on the Areca. Another advantage of using the Areca is that you not fill up your system as much as when adding the 4 port card, thus leaving more room for airflow and effective cooling. If the 1231 has ML (multi-lane) connections, all you need is a single forward breakout cable from the Areca to the eSATA drives (up to 4). |
June 3rd, 2009, 10:10 AM | #12 |
Inner Circle
Join Date: Jan 2007
Location: Woodinville, WA USA
Posts: 3,467
|
Thanks Harm. I knew I could count on you.
I may do a little experimenting when my current projects are done in a few weeks. I'll report back if I find anything interesting. |
June 5th, 2009, 12:56 PM | #13 |
Inner Circle
Join Date: Jan 2007
Location: Woodinville, WA USA
Posts: 3,467
|
So, as is usual whenever Harm comes up with a really intriguing suggestion, I've gotten myself in over my head. Well, I haven't actually done anything yet, so therefore have not destroyed my system, but after opening up the SuperMicro case I now have many questions. This is still all about eSATA drives, so I hope it's not too OT. But here goes:
The Areca 1231ML controller has three SFF-8087 ports on the card, according to the manual. --First of all, every reference to SFF-8087 I can find calls it SAS. Do SATA and SAS use the same connector? --Of the three ports, two are connected to bundles of red wires, 7 of which go to the backplane and one of which dangles, unused. The third port remains unoccupied. My thought was to connect the breakout cable you suggested to this unoccupied port. But all I can find are SAS to SATA x4. Is there such a thing as SFF-8087 to 4x eSATA? Or should I buy four adapters after it splits? And is there a length limit? Finally, the mobo has four unoccupied SATA ports. If my external eSATA drives are just going to be JBODs, is there a performance penalty just coming off the mobo rather than the Areca? And is here where I could just use SATA to eSATA connectors? Sorry for so many esoteric questions... it's always dangerous when I get to thinking. The only reason this matters at all is that I'm out of slots in the case and would have to sacrifice my FW add-in card to make room for the eSATA add-in card. Any thoughts or advice here would be most appreciated. |
June 5th, 2009, 01:45 PM | #14 | |
Trustee
Join Date: Aug 2006
Location: Rotterdam, Netherlands
Posts: 1,832
|
Quote:
When you use, as I suggested, a single multi-lane 8087 forward breakout cable, you end up with a single connector on the Areca and 4 SATA connectors for your external disks. The normal length of these cables is around 32" but you may be able to get longer ones. That still leaves you with the power connection requirements, that are circumvented with SAS connectors, that include both SATA and power. But unfortunatley that is not possible with your controller card. So what you need is an external housing with its own power supply, cooling etc. and the forward 8087 breakout cable. (BTW, Areca cables may be difficult to find, but 3Ware uses the same ones and are more readily available). Adam, there will likely be a slight, and I really mean slight (at least in your case for your purposes) performance penalty if you connect the eSATA disks to your on-board controller. However, that does not solve the connection and power supply issue. You still need to feed them with some form of juice and you need SATA connections. I still think the easiest way out is to use the ML capability of the 1231. |
|
June 5th, 2009, 02:49 PM | #15 |
Major Player
Join Date: Dec 2006
Location: Austin, TX
Posts: 773
|
I tried using ESATA on my MacBook Pro, but I kept getting kernel panics.
I use it on my editing rig in Vista and it works fine - it's great. ESata is basically like having access to an "internal" hard drive from outside your box. Don't really see much of a reason to use anything else if you've got it.
__________________
Equip: Panny GH1, Canon HG20, Juicedlink, AT897, Sennh. EW/GW100, Zoom H2, Vegas 8.1 |
| ||||||
|
|