4 ms·

USB drives with no phantom load

56 points by edward 8y ago

comboy 8y ago
It's cool but I don't know if I would want additional latency (especially in the case where they are currently going to sleep) plus possibility of turning them off at the wrong moment to save $0.2 per month.
- Someone 8y ago
  Why do you think this is about cost? The article says ”I've not had any network attached storage at home, because it's offgrid”, so I think it’s about preventing the house from powering down.
  - pronoiac 8y ago
    I can't find a concise page to link, but he has a house that's off-the-grid, using solar panels and batteries. So he pays attention to power consumption.
    - bcaa7f3a8bbc 8y ago
      Link this page. https://lwn.net/Articles/672352/ https://lwn.net/Articles/672352/
      - lkj 8y ago
        Thank you! The following quote resonates heavily with my mind: > I'm fairly addicted to that point in development of a project where it's all about exploring a vast solution space, and making countless little choices that will hopefully add up to something coherent and well thought out and useful. Or might fail gloriously.
jws 8y ago
I share the author’s concerns for off grid power use. 3 watts per drive (I assume a wall wart for a 3.5” drive) works out to 72watt hours per day. About 1/10th of a $200 solar panel. Double or triple that to cover batteries and electronics. Then you run into discontinuities in system sizing where you can’t easily add just one more panel. Conserving energy suddenly makes a lot of sense. USB hubs with per port power control are the flying unicorn of hardware. Once in a while some vendor slips up and makes one, but by the time people figure it out the vendor has ended the model or replaced it with a different model under the same model number. It’s especially frustrating because many of the USB hub chipsets support the feature, but only if the manufacturer puts a switching transistor on each port. It literally would only cost pennies per port to have the feature.
- nexuist 8y ago
  Why do vendors not want per port power control?
  - DarkStar851 8y ago
    It's not that they don't want to, it's just a lot of extra cost for a product their bulk userbase won't be all that interested in. Think about it, before today (or even after), would you have spent $20+ more for a USB hub with per-port power control?
    - OJFord 8y ago
      I got one a couple of weeks ago for under 10. Didn't realise it was at all odd or rare.
      - OJFord 8y ago
        Ah - sorry - I got an external hub; read this initially as I was waking up and completely misunderstood.
    - makomk 8y ago
      Not that much extra cost, I don't think. MOSFETs capable of switching USB loads are in the low single-digit cents each. It's probably just that everyone wants to squeeze out the last few cents of manufacturing cost.
Abishek_Muthian 8y ago
Tangentially, I'm curious on how OP monitors health of USB drives. Getting SMART data from USB drives has been a pain on linux as the kernel rejects SAT ATA pass-through commands & so doesn't work with smartmontools[1]. I have tested across different USB 3.0 drives such as Seagate Backup Plus, MYDIGITALSSD|OTG, SAMSUNG Portable SSD T5 all of which use 'uas' driver under linux & if used with USB 3.0 hub like Tplink USB 3.0 port hub; it bridges using 'hub/4p' driver. After trying out several other methods in vain, the only tool which I found to be helpful in checking for USB drive health on linux is discscan[2]; which tests for unreadable sectors & latency without using SMART values. [1]: https://www.smartmontools.org/wiki/USB https://www.smartmontools.org/wiki/USB [2]: https://github.com/baruch/diskscan https://github.com/baruch/diskscan
- Vogtinator 8y ago
  I have never any issues with USB HDDs with smartctl. In the rare cases where autodetection did not work, smartctl -d <type> with trial and error works.
  - Abishek_Muthian 8y ago
    I see, I forgot to mention - d scsi gives only basic manufacturer info and tests fail. Does your drive work in uas mode?
zokier 8y ago
3 W per drive when shut down seems like huge amount. For reference, Raspberry Pi 3B+ consumes about 2 W when idle, and 2B 2.5 W under load. https://www.pidramble.com/wiki/benchmarks/power-consumption https://www.pidramble.com/wiki/benchmarks/power-consumption

56 points by edward 8y ago

It's cool but I don't know if I would want additional latency (especially in the case where they are currently going to sleep) plus possibility of turning them off at the wrong moment to save $0.2 per month.

Someone 8y ago
Why do you think this is about cost? The article says ”I've not had any network attached storage at home, because it's offgrid”, so I think it’s about preventing the house from powering down.
- pronoiac 8y ago
  I can't find a concise page to link, but he has a house that's off-the-grid, using solar panels and batteries. So he pays attention to power consumption.
  - bcaa7f3a8bbc 8y ago
    Link this page. https://lwn.net/Articles/672352/ https://lwn.net/Articles/672352/
    - lkj 8y ago
      Thank you! The following quote resonates heavily with my mind: > I'm fairly addicted to that point in development of a project where it's all about exploring a vast solution space, and making countless little choices that will hopefully add up to something coherent and well thought out and useful. Or might fail gloriously.

jws 8y ago

I share the author’s concerns for off grid power use. 3 watts per drive (I assume a wall wart for a 3.5” drive) works out to 72watt hours per day. About 1/10th of a $200 solar panel. Double or triple that to cover batteries and electronics. Then you run into discontinuities in system sizing where you can’t easily add just one more panel. Conserving energy suddenly makes a lot of sense. USB hubs with per port power control are the flying unicorn of hardware. Once in a while some vendor slips up and makes one, but by the time people figure it out the vendor has ended the model or replaced it with a different model under the same model number. It’s especially frustrating because many of the USB hub chipsets support the feature, but only if the manufacturer puts a switching transistor on each port. It literally would only cost pennies per port to have the feature.

nexuist 8y ago
Why do vendors not want per port power control?
- DarkStar851 8y ago
  It's not that they don't want to, it's just a lot of extra cost for a product their bulk userbase won't be all that interested in. Think about it, before today (or even after), would you have spent $20+ more for a USB hub with per-port power control?
  - OJFord 8y ago
    I got one a couple of weeks ago for under 10. Didn't realise it was at all odd or rare.
    - OJFord 8y ago
      Ah - sorry - I got an external hub; read this initially as I was waking up and completely misunderstood.
  - makomk 8y ago
    Not that much extra cost, I don't think. MOSFETs capable of switching USB loads are in the low single-digit cents each. It's probably just that everyone wants to squeeze out the last few cents of manufacturing cost.

Abishek_Muthian 8y ago

Tangentially, I'm curious on how OP monitors health of USB drives. Getting SMART data from USB drives has been a pain on linux as the kernel rejects SAT ATA pass-through commands & so doesn't work with smartmontools[1]. I have tested across different USB 3.0 drives such as Seagate Backup Plus, MYDIGITALSSD|OTG, SAMSUNG Portable SSD T5 all of which use 'uas' driver under linux & if used with USB 3.0 hub like Tplink USB 3.0 port hub; it bridges using 'hub/4p' driver. After trying out several other methods in vain, the only tool which I found to be helpful in checking for USB drive health on linux is discscan[2]; which tests for unreadable sectors & latency without using SMART values. [1]: https://www.smartmontools.org/wiki/USB https://www.smartmontools.org/wiki/USB [2]: https://github.com/baruch/diskscan https://github.com/baruch/diskscan

Vogtinator 8y ago
I have never any issues with USB HDDs with smartctl. In the rare cases where autodetection did not work, smartctl -d <type> with trial and error works.
- Abishek_Muthian 8y ago
  I see, I forgot to mention - d scsi gives only basic manufacturer info and tests fail. Does your drive work in uas mode?

zokier 8y ago

3 W per drive when shut down seems like huge amount. For reference, Raspberry Pi 3B+ consumes about 2 W when idle, and 2B 2.5 W under load. https://www.pidramble.com/wiki/benchmarks/power-consumption https://www.pidramble.com/wiki/benchmarks/power-consumption