LINUX LITE 7.2 FINAL RELEASED - SEE RELEASE ANNOUNCEMENTS SECTION FOR DETAILS


Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
How do I combine several small HDDs to make one large HDD?
#11
I am unsure but I think other factors could be ?

Each file is spread across the 4 hard drives, it is why if one fails the file cannot be fully recovered, does this mean each of the smaller hard drives work less for a file of a given size in comparison to the single large hard drive for a file of the same given size ?

1. on each of the spanned 4 drives -
the arm moves less often and the disk spins less, the clusters used are smaller as the given file is distributed over the 4 in the span (ultimately meaning increased movement, does this yet still = the larger hdd stresses)?

2. Does each individual psu work less whilst spanning and locating due to these factors ?
Is the total of voltage and current drawn for the reduced actions of the four smaller hard drives <,=,> that of the larger single hard drive that works harder for the given large file, in fact will it use the same size psu to begin with, or a scaling be used to calculate this?

3. to determine the factor of heat on the psu whilst working, if it is the same size as the 4 smaller hard drives (or even not) the single large hard drive psu works harder and gets hotter, in theory the larger hard drive psu will fail sooner, due to the effect of heat. If all psu were of the same quality to begin with ?

4. the hard drives are working in a series circuit, like 4 light bulbs in such if one fails they all turn off. If it were possible a radial circuit would be better, then this would not happen as one failing would not have any effect on the others in the circuit.
Radial then it would not be spanned.<< Instead somehow to use switching of multiple hdd that is automated, so the total volume is used/combined, but the individual file is not spread across more than a single hdd, thus if one hdd fails the data on the other 3 is fine, it may also be possible to recover files from the failed hdd in some instances.
That as space is used and filled it begins on the next hdd, if it were intelligent design, it could decide 1 large file would be too much for a remaining capacity of a hdd, and put it on the next hdd, then a later file could fit in the first hdd final capacity so it saves it there, if it is possible ?

5. The power supplies are not in a parallel circuit but their effect on the spanned hard drives are working in the similar manner ?
eg; in that the 4 hard drives overall as a series unit, do receive the same total voltage and could draw the same rate of current (if all psu and hard drives were the same), but not from a single source, as there are 4 psu being used, so the needed parameters are spread across 4 psu each having less load on them and less heat produced.
That is each hard drive has the smallest and largest possibility of the psu available to it.
v/s
The single large hdd psu will work harder and get hotter, and will fail sooner than any of the 4 psu of the spanned hdd ?

6. If 1 thru 5 are measured, calculated, then the probabilities of 4 hard drives v/s 1 can be used like so -
there will be a "sweet number", that is a number of spanned hard drives that prove to be more reliable than a single hard drive of a given total volume, where all numbers of spanned hard drives below that have lower success rates compared to a single large hard drive, and a range above the sweet number it has no improved results over a single hard drive, then upwards from that reaching a final number where the spanned hard drives become more and more unreliable per hard drive added when compared to the single large volume, or maybe not ? Wink
That result would give the best total hard drive size spanned v/s it's equivalent single hard drive.

Update -

I just read this link "Option Three - RAID Array" - https://lifehacker.com/5986883/how-to-co...ty-storage

RAID 10 or RAID 5 looks to solve the issue of hdd failure, I'm not sure how the system is built, I'll look later for some info on it.
Reply
#12
This statement appears towards the end of the document  "Option Three - RAID Array" which you quoted.

"Cons: Unfortunately, RAID has a few downsides too. To start, higher levels of RAID can be expensive, since you need quite a few drives to pull it off. They'd also have to be the same size and speed, or you'd have to sacrifice some of their size and speed. That is, in a RAID array, you're stuck using the smallest disk's capacity for each drive, and the slowest disk's speed for each drive. As such, it isn't ideal if you have a bunch of disks lying around and you're trying to save some money. But, if your data is important, it can help keep that data safe, too. RAID 0's biggest downside is not only the lack of redundancy, but the fact that if you lose one drive, you lose all your data—with little to no hope of recovering it."

Not worth considering IMHO - just convert the drives to USB and connect as required. I'm sure you don't need
multi terrabytes of data all at once. Easier to backup too.  Wink
1) Lenovo T520 i5 LL3.8 8GB ram, fast & stable
2) Medion P4 32bit LL3.8 1GB ram, quite fast & stable
3) eeePC 901 32bit LL3.8 1GB ram, fast & stable
4) eeePC 701 32bit LL3.8 1GB ram, slower & stable but small and light enough to travel with me to New Zealand when visiting family in Blenheim.
Reply
#13
(01-14-2018, 04:13 PM)justme2 link Wrote: This statement appears towards the end of the document  "Option Three - RAID Array" which you quoted.

"Cons: Unfortunately, RAID has a few downsides too. To start, higher levels of RAID can be expensive, since you need quite a few drives to pull it off. They'd also have to be the same size and speed, or you'd have to sacrifice some of their size and speed. That is, in a RAID array, you're stuck using the smallest disk's capacity for each drive, and the slowest disk's speed for each drive. As such, it isn't ideal if you have a bunch of disks lying around and you're trying to save some money. But, if your data is important, it can help keep that data safe, too. RAID 0's biggest downside is not only the lack of redundancy, but the fact that if you lose one drive, you lose all your data—with little to no hope of recovering it."

Not worth considering IMHO - just convert the drives to USB and connect as required. I'm sure you don't need
multi terrabytes of data all at once. Easier to backup too.  Wink

Many thanks for your common sense overview.  I think with all that has been said here on this thread, I'm fast coming to the conclusion it might be simpler, easier & more secure just to buy & use a single larger USB drive for all my video data which is now around 3TB. I have a 4TB USB Expansion Portable Drive (3.6TB formatted drivespace), which I have been very pleased with. A 5TB one has come out, but I might sit tight until an 8TB one comes out, though I think I might be waiting for a while ...

Thanks everybody for your interest in this thread and for the interesting & useful info & links supplied.

Mike 
64bit OS (32-bit on Samsung[i] netbook) installed in [i]Legacy mode on MBR-formatted SSDs (except pi which uses a micro SDHC card):
2017 - Raspberry pi 3B (4cores) ~ [email protected] - LibreElec, used for upgrading our Samsung TV (excellent for the task)  
2012 - Lenovo G580 2689 (2cores; 4threads] ~ [email protected] - LL3.8/Win8.1 dual-boot (LL working smoothly)
2011 - Samsung NP-N145 Plus (1core; 2threads) ~ Intel Atom [email protected] - LL 3.8 32-bit (64-bit too 'laggy')
2008 - Asus X71Q (2cores) ~ Intel [email protected] - LL4.6/Win8.1 dual-boot, LL works fine with kernel 4.15
2007 - Dell Latitude D630 (2cores) ~ Intel [email protected] - LL4.6, works well with kernel 4.4; 4.15 doesn't work
Reply
#14
(01-14-2018, 01:32 PM)bitsnpcs link Wrote: I am unsure but I think other factors could be ?

Each file is spread across the 4 hard drives, it is why if one fails the file cannot be fully recovered, does this mean each of the smaller hard drives work less for a file of a given size in comparison to the single large hard drive for a file of the same given size ?

1. on each of the spanned 4 drives -
the arm moves less often and the disk spins less, the clusters used are smaller as the given file is distributed over the 4 in the span (ultimately meaning increased movement, does this yet still = the larger hdd stresses)?

2. Does each individual psu work less whilst spanning and locating due to these factors ?
Is the total of voltage and current drawn for the reduced actions of the four smaller hard drives <,=,> that of the larger single hard drive that works harder for the given large file, in fact will it use the same size psu to begin with, or a scaling be used to calculate this?

3. to determine the factor of heat on the psu whilst working, if it is the same size as the 4 smaller hard drives (or even not) the single large hard drive psu works harder and gets hotter, in theory the larger hard drive psu will fail sooner, due to the effect of heat. If all psu were of the same quality to begin with ?

4. the hard drives are working in a series circuit, like 4 light bulbs in such if one fails they all turn off. If it were possible a radial circuit would be better, then this would not happen as one failing would not have any effect on the others in the circuit.
Radial then it would not be spanned.<< Instead somehow to use switching of multiple hdd that is automated, so the total volume is used/combined, but the individual file is not spread across more than a single hdd, thus if one hdd fails the data on the other 3 is fine, it may also be possible to recover files from the failed hdd in some instances.
That as space is used and filled it begins on the next hdd, if it were intelligent design, it could decide 1 large file would be too much for a remaining capacity of a hdd, and put it on the next hdd, then a later file could fit in the first hdd final capacity so it saves it there, if it is possible ?

5. The power supplies are not in a parallel circuit but their effect on the spanned hard drives are working in the similar manner ?
eg; in that the 4 hard drives overall as a series unit, do receive the same total voltage and could draw the same rate of current (if all psu and hard drives were the same), but not from a single source, as there are 4 psu being used, so the needed parameters are spread across 4 psu each having less load on them and less heat produced.
That is each hard drive has the smallest and largest possibility of the psu available to it.
v/s
The single large hdd psu will work harder and get hotter, and will fail sooner than any of the 4 psu of the spanned hdd ?

6. If 1 thru 5 are measured, calculated, then the probabilities of 4 hard drives v/s 1 can be used like so -
there will be a "sweet number", that is a number of spanned hard drives that prove to be more reliable than a single hard drive of a given total volume, where all numbers of spanned hard drives below that have lower success rates compared to a single large hard drive, and a range above the sweet number it has no improved results over a single hard drive, then upwards from that reaching a final number where the spanned hard drives become more and more unreliable per hard drive added when compared to the single large volume, or maybe not ? Wink
That result would give the best total hard drive size spanned v/s it's equivalent single hard drive.

Update -

I just read this link "Option Three - RAID Array" - https://lifehacker.com/5986883/how-to-co...ty-storage

RAID 10 or RAID 5 looks to solve the issue of hdd failure, I'm not sure how the system is built, I'll look later for some info on it.

Wow, some great work here bitsnpcs - thankyou so much for your thoughts and effort you put into your reply. Really appreciated

Mike
64bit OS (32-bit on Samsung[i] netbook) installed in [i]Legacy mode on MBR-formatted SSDs (except pi which uses a micro SDHC card):
2017 - Raspberry pi 3B (4cores) ~ [email protected] - LibreElec, used for upgrading our Samsung TV (excellent for the task)  
2012 - Lenovo G580 2689 (2cores; 4threads] ~ [email protected] - LL3.8/Win8.1 dual-boot (LL working smoothly)
2011 - Samsung NP-N145 Plus (1core; 2threads) ~ Intel Atom [email protected] - LL 3.8 32-bit (64-bit too 'laggy')
2008 - Asus X71Q (2cores) ~ Intel [email protected] - LL4.6/Win8.1 dual-boot, LL works fine with kernel 4.15
2007 - Dell Latitude D630 (2cores) ~ Intel [email protected] - LL4.6, works well with kernel 4.4; 4.15 doesn't work
Reply
#15
(01-14-2018, 04:13 PM)justme2 link Wrote: This statement appears towards the end of the document  "Option Three - RAID Array" which you quoted.

"Cons: Unfortunately, RAID has a few downsides too. To start, higher levels of RAID can be expensive, since you need quite a few drives to pull it off. They'd also have to be the same size and speed, or you'd have to sacrifice some of their size and speed. That is, in a RAID array, you're stuck using the smallest disk's capacity for each drive, and the slowest disk's speed for each drive. As such, it isn't ideal if you have a bunch of disks lying around and you're trying to save some money. But, if your data is important, it can help keep that data safe, too. RAID 0's biggest downside is not only the lack of redundancy, but the fact that if you lose one drive, you lose all your data—with little to no hope of recovering it."

Not worth considering IMHO - just convert the drives to USB and connect as required. I'm sure you don't need
multi terrabytes of data all at once. Easier to backup too.  Wink

No I don't need much space at all.
Reply
#16
(01-14-2018, 06:09 PM)m654321 link Wrote: Wow, some great work here bitsnpcs - thankyou so much for your thoughts and effort you put into your reply. Really appreciated

Mike

I enjoyed your thread it was interesting for me seeing/reading lots of new things/possibilities whilst looking around about things.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)