08-04-2018, 09:29 AM
(This post was last modified: 08-05-2018, 04:09 AM by jovval.
Edit Reason: I elaborated ending.
)
TAKE-AWAY: tkaiser's findings (shown last, at the end of #3) with insufficient power may relate to the wacky results found by the OP. Has anyone reproduced the errors with a 5A PSU or powered hub with more amperage?
I elaborate on my preceding post with this supplement. I believe #2 (below) has the "very good and long thread from tkaiser on Armbian forum where he describes SATA bridges, NAS solutions, and Rock64 performance numbers and he has good explanations what setups will work correctly" mentioned in #2 of my previous post. But. The link has performance numbers yet lacks or has in underdeveloped form (for my purposes) NAS solutions and SATA bridges.
Please note my respectful differences with tkaiser, and correct me if I am underinformed or just plain wrong. Hey. It happens, you know?
(1) Rock64 Mini NAS and media center, works? - https://forum.armbian.com/topic/4958-roc...ter-works/
By stefan.z.j, August 20, 2017 in Rockchip 3288 & 3328
Pertinant posts follow.
NAS backed by a USB3 Raid5 enclosure
[tkaiser responds]
IMO a really bad idea. Those proprietary RAID boxes are just a huge single point of failure and I would never rely on such things (but I do storage for a living and only deal with RAID any more when it failed so I'm a bit biased here).
Wrt performance: ROCK64 is limited by Gigabit Ethernet, real world throughput with Windows Explorer or macOS Finder will be around or even above 100 MB/s (megabyte) given the USB3 attached storage is fast enough.
[tkaiser's point is well-taken, but better phrased in (a). I address shortcomings in (b-e).
(a) The RAID controller is a black box, mechanisms known only to the illuminati. If it fails, who knows what befalls your data.
(b) You can have two controllers, and swap out the bad one for the good one. If the latter works, you still depend on the bad one not having corrupted your data.
© Local backups address most of the deficiencies (ideally having redundant redundancy, i.e. two RAIDs).
(d) Remote backups address the problem more completely (with a pair of remote RAIDs, naturally, each at a different location).
(e) If cost concerns you, you can still consider non-RAID backups: (i) on and off-line, (ii) local and remote.
(f) I have no statistics to support my statements. Be sure to get a second and third opinion of (a-e).]
[tkaiser continues]
I would suggest not using RAID at home since it sucks (it works as long as you won't need it but once a disk fails funny things happen like the whole array disappearing). [These "funny things" are exactly the reason why I need to have a controller tell me when a disk is failing, or about to fail. If the controller will not do so, software RAID with JBOD should suffice, and may be more readily transferrable to different equipment.] Also the redudant RAID modes are not about data protection/safety but only about availability (who needs this at home?). [Those with home offices. So does anyone who does not want to come home from work to find more work. Maybe even you.]
(2) Some storage benchmarks on SBCs - https://forum.armbian.com/topic/1925-som...s-on-sbcs/
By tkaiser, August 30, 2016 in Research guides & tutorials [USB 3.0 mentioned 6x; Rock64, 17x beginning June 22, 2017. Last post, 5/10/2018]
Posted June 22, 2017 (edited)
All tests done with Samsung EVO SSDs (my EVO840 used for all tests [begin strike-out]except of the ODROID-XU4 results which were made with a much faster EVO850 instead and[end strike-out] Pine64 numbers with a slower EVO750 so random IO numbers should be multiplied with 1.3 or even 1.4):
Random IO in IOPS Sequential IO in MB/sec
4K read/write 1M read/write
Pine64 (USB2/UAS) 2260/1948 42 / 41
Banana Pi Pro (SATA) 3943/3478 122 / 37
Wandboard Quad (SATA) 4141/5073 100 / 100
ODROID-XU4 (USB3/UAS) 4637/5126 262 / 282
ROCK64 (USB3/UAS) 4619/5972 311 / 297
Clearfog Pro (SATA) 10148/19184 507 / 448
USB3 SBC like ODROID-XU4/HC1 or ROCK64 show way better performance. You only have to take care of the USB downsides (XU4 with an internal USB hub and receptacles problems is in a bad position here) and ensure that you're using good USB-to-SATA bridges to connect SSD or HDD
If we also take price/performance ratio into account then ROCK64 or other RK3328 boards that will follow are really hard to beat. The 1GB ROCK64 variant will most probably stay below the $25 margin and you should keep in mind that more DRAM is pretty useless if you think about (NAS) performance.
tkaiser, Posted March 19, Early 2018 update [Contains and expands upon the data above. I include only new outperformers of Rock64.]
Random IO in IOPS Sequential IO in MB/sec
4K read/write 1M read/write
RK3399 (USB3/UAS) 5994/6308 330 / 340
ASM1061 powersave 6010/7900 320 / 325
ASM1061 performance 9820/16230 330 / 330
RK3399-Q7 (NVMe) 11640/36900 1070 / 1150
(3) [Mini review] ROCK64 SATA cable - https://forum.armbian.com/topic/5098-min...ata-cable/
[All of you know or /should know/ the content of this link.
Check. it. out.
It's concluding paragraph puzzles me.]
The 5V power requirements for JMS578 + 3.5" HDD disk were as follows:
USB2: 6.4W vs. 3.3W (full load vs. idle). Numbers with 5V PSU included so we're talking about needed power provided on an USB2 port of approximately ~3W which fits perfectly in the power budget of ROCK64's USB2: 650mA * 5V == 3250mW
USB3: 7.9W vs. 3.3W (full load vs. idle). Numbers again with 5V PSU included so we're talking about needed power provided on an USB3 port of approximately ~4W which fits perfectly in the power budget of ROCK64's USB3: 950mA * 5V == 4750mW
At least with my 3.5" HDD it worked pretty well to let it operate on both USB2 and USB3 ports when board powering was appropriate (with insufficient powering all weird sorts of issues popped up. My favourite was a kernel panic when issueing 'lsusb' after 30 seconds. Once I powered ROCK64 reliably all these 'software issues' were gone immediately -- again and again: insufficient powering is one of the most severe problem sources)
End of forums and their excepts. My concluding thoughts.
The clause "with insufficient powering all weird sorts of issues popped up" begs the question: what's he talking about?
More succinctly, is the 3A power block from the Pine store insufficient?
Is the Rock64Pro 5A power block compatible with the Rock64? If not, what alternative do you suggest?
Also, the sputtering of several SATA drives may stem from a need for a more powerful USB hub. I know that even 2A /should/ cover 2 x 4W (( 2 x 4W) / 5V = 1.6A) that tkaiser attributes to his 3.5" drives' use of 5V, but I am still wondering if, in practice, 2A does. Clearly, it can not handle more than 2 3.5" SATA HDDs. And I am presuming that the powered hub is functioning as it should, but it may not be. Nor may the power supply be up to snuff. (Yes, I know the OP was using 2.5" drives. I am just being thorough.)
I would like to know the output of the hubs used by the OP and any others experiencing problems with multiple SATA drives connected via USB.
TAKE-AWAY: tkaiser's findings with insufficient power may relate to the wacky results found by the OP. Has anyone reproduced the errors with a 5A PSU or powered hub with more amperage?
Over and out.
I elaborate on my preceding post with this supplement. I believe #2 (below) has the "very good and long thread from tkaiser on Armbian forum where he describes SATA bridges, NAS solutions, and Rock64 performance numbers and he has good explanations what setups will work correctly" mentioned in #2 of my previous post. But. The link has performance numbers yet lacks or has in underdeveloped form (for my purposes) NAS solutions and SATA bridges.
Please note my respectful differences with tkaiser, and correct me if I am underinformed or just plain wrong. Hey. It happens, you know?
(1) Rock64 Mini NAS and media center, works? - https://forum.armbian.com/topic/4958-roc...ter-works/
By stefan.z.j, August 20, 2017 in Rockchip 3288 & 3328
Pertinant posts follow.
NAS backed by a USB3 Raid5 enclosure
[tkaiser responds]
IMO a really bad idea. Those proprietary RAID boxes are just a huge single point of failure and I would never rely on such things (but I do storage for a living and only deal with RAID any more when it failed so I'm a bit biased here).
Wrt performance: ROCK64 is limited by Gigabit Ethernet, real world throughput with Windows Explorer or macOS Finder will be around or even above 100 MB/s (megabyte) given the USB3 attached storage is fast enough.
[tkaiser's point is well-taken, but better phrased in (a). I address shortcomings in (b-e).
(a) The RAID controller is a black box, mechanisms known only to the illuminati. If it fails, who knows what befalls your data.
(b) You can have two controllers, and swap out the bad one for the good one. If the latter works, you still depend on the bad one not having corrupted your data.
© Local backups address most of the deficiencies (ideally having redundant redundancy, i.e. two RAIDs).
(d) Remote backups address the problem more completely (with a pair of remote RAIDs, naturally, each at a different location).
(e) If cost concerns you, you can still consider non-RAID backups: (i) on and off-line, (ii) local and remote.
(f) I have no statistics to support my statements. Be sure to get a second and third opinion of (a-e).]
[tkaiser continues]
I would suggest not using RAID at home since it sucks (it works as long as you won't need it but once a disk fails funny things happen like the whole array disappearing). [These "funny things" are exactly the reason why I need to have a controller tell me when a disk is failing, or about to fail. If the controller will not do so, software RAID with JBOD should suffice, and may be more readily transferrable to different equipment.] Also the redudant RAID modes are not about data protection/safety but only about availability (who needs this at home?). [Those with home offices. So does anyone who does not want to come home from work to find more work. Maybe even you.]
(2) Some storage benchmarks on SBCs - https://forum.armbian.com/topic/1925-som...s-on-sbcs/
By tkaiser, August 30, 2016 in Research guides & tutorials [USB 3.0 mentioned 6x; Rock64, 17x beginning June 22, 2017. Last post, 5/10/2018]
Posted June 22, 2017 (edited)
All tests done with Samsung EVO SSDs (my EVO840 used for all tests [begin strike-out]except of the ODROID-XU4 results which were made with a much faster EVO850 instead and[end strike-out] Pine64 numbers with a slower EVO750 so random IO numbers should be multiplied with 1.3 or even 1.4):
Random IO in IOPS Sequential IO in MB/sec
4K read/write 1M read/write
Pine64 (USB2/UAS) 2260/1948 42 / 41
Banana Pi Pro (SATA) 3943/3478 122 / 37
Wandboard Quad (SATA) 4141/5073 100 / 100
ODROID-XU4 (USB3/UAS) 4637/5126 262 / 282
ROCK64 (USB3/UAS) 4619/5972 311 / 297
Clearfog Pro (SATA) 10148/19184 507 / 448
USB3 SBC like ODROID-XU4/HC1 or ROCK64 show way better performance. You only have to take care of the USB downsides (XU4 with an internal USB hub and receptacles problems is in a bad position here) and ensure that you're using good USB-to-SATA bridges to connect SSD or HDD
If we also take price/performance ratio into account then ROCK64 or other RK3328 boards that will follow are really hard to beat. The 1GB ROCK64 variant will most probably stay below the $25 margin and you should keep in mind that more DRAM is pretty useless if you think about (NAS) performance.
tkaiser, Posted March 19, Early 2018 update [Contains and expands upon the data above. I include only new outperformers of Rock64.]
Random IO in IOPS Sequential IO in MB/sec
4K read/write 1M read/write
RK3399 (USB3/UAS) 5994/6308 330 / 340
ASM1061 powersave 6010/7900 320 / 325
ASM1061 performance 9820/16230 330 / 330
RK3399-Q7 (NVMe) 11640/36900 1070 / 1150
(3) [Mini review] ROCK64 SATA cable - https://forum.armbian.com/topic/5098-min...ata-cable/
[All of you know or /should know/ the content of this link.
Check. it. out.
It's concluding paragraph puzzles me.]
The 5V power requirements for JMS578 + 3.5" HDD disk were as follows:
USB2: 6.4W vs. 3.3W (full load vs. idle). Numbers with 5V PSU included so we're talking about needed power provided on an USB2 port of approximately ~3W which fits perfectly in the power budget of ROCK64's USB2: 650mA * 5V == 3250mW
USB3: 7.9W vs. 3.3W (full load vs. idle). Numbers again with 5V PSU included so we're talking about needed power provided on an USB3 port of approximately ~4W which fits perfectly in the power budget of ROCK64's USB3: 950mA * 5V == 4750mW
At least with my 3.5" HDD it worked pretty well to let it operate on both USB2 and USB3 ports when board powering was appropriate (with insufficient powering all weird sorts of issues popped up. My favourite was a kernel panic when issueing 'lsusb' after 30 seconds. Once I powered ROCK64 reliably all these 'software issues' were gone immediately -- again and again: insufficient powering is one of the most severe problem sources)
End of forums and their excepts. My concluding thoughts.
The clause "with insufficient powering all weird sorts of issues popped up" begs the question: what's he talking about?
More succinctly, is the 3A power block from the Pine store insufficient?
Is the Rock64Pro 5A power block compatible with the Rock64? If not, what alternative do you suggest?
Also, the sputtering of several SATA drives may stem from a need for a more powerful USB hub. I know that even 2A /should/ cover 2 x 4W (( 2 x 4W) / 5V = 1.6A) that tkaiser attributes to his 3.5" drives' use of 5V, but I am still wondering if, in practice, 2A does. Clearly, it can not handle more than 2 3.5" SATA HDDs. And I am presuming that the powered hub is functioning as it should, but it may not be. Nor may the power supply be up to snuff. (Yes, I know the OP was using 2.5" drives. I am just being thorough.)
I would like to know the output of the hubs used by the OP and any others experiencing problems with multiple SATA drives connected via USB.
TAKE-AWAY: tkaiser's findings with insufficient power may relate to the wacky results found by the OP. Has anyone reproduced the errors with a 5A PSU or powered hub with more amperage?
Over and out.