NAS Drives

[Little Jim]

I would like to get a NAS drive to use at home.  It will mainly be for storing music but perhaps we'll use it to back up stuff like photos on as well.  We have recently bought a Sonos so it will be mainly used for feeding that with music.  So the question is what size discs should I get?

Am I right in thinking that the maximum a music CD can store is just over 700 MB, so 0.7GB and a 1 TB disc will store 1000 x as much as a 1 GB disc, so a bit over 1000 (completely full) CDs ripped using something like FLAC?  or am I getting my 10x table in a twist?  So, in other words will a 2 TB disc be more than enough storage capacity (except that I know that you can never have too much storage capacity!)

[Kim]

A back of the envelope calculation based on my CDs-ripped-as-FLAC collection yields an average of 331MB per disc.  That's mostly albums (the few singles and compilations cancel out, size-wise).  A CD does indeed store about 700MB of PCM audio, but FLAC compresses that to a little over half the size, and most discs^[1] don't use all the available capacity.  (Obviously you can achieve a much higher compression ratio using lossy codecs.)

To muddy the waters, computers - operating in binary - calculate file and memory sizes in powers of 2.  So 1TiB = 1024GiB = 1048576MiB.  Disk^[1] and network equipment manufacturers, on the other hand, tend to use decimal SI prefixes, so 1TB = 1000GB = 1000000MB.  This makes some sense for telecoms, but is mostly about making the numbers sound slightly more impressive when it comes to storage.  (The IEC coined the infrequently used terms 'tebibyte' 'gibibyte' 'mebibyte' for the binary prefixes in a mostly-futile attempt at clarification.)

In the real world, we tend to handwave over the difference, and just bear in mind that a given volume will have a few percent less usable capacity after the filesystem overheads and conversion to binary prefixes.


My rule of thumb for buying storage for domestic use is to work out how much you're probably going to need in 2-3 years time (the expected lifetime of hard disks), and buy whichever size has the best storage:price ratio above that.


_{[1] By convention, optical discs are spelt with a 'c' and magnetic disks are spelt with a 'k'.}

[Little Jim]

Kim, thank you very much for a reply that I can understand!

So (give or take) that is about 3 CDs per 1 GB, or 3000 on a 1 TB disc (without making any allowance for the OS to run the NAS drive) so a pair of 2 TB discs should be more than enough even allowing for various other things that it seems like a good-idea-at-the-time get put on there as well.

[ian]

Don't forget backups, otherwise that might be a lot of disks to re-rip...

[Kim]

Don't forget backups, otherwise that might be a lot of disks to re-rip...

My take on that is that the CDs are the backup.

Well, that's not actually true, as our ~42GiB of CD rips is small enough to get gobbled up by my usual backup strategy, but it's one of the things that I wouldn't loose too much sleep over if I lost the data.  It was only an evening swapping discs (admittedly because I was able to parallelise about 6 CD drives across 4 computers).

[ian]

Partly why I used MP3 and AAC for my collection (I confess I now buy digital), a couple of thousand disks run to a fraction of a terabyte, easy enough for back-ups both locally and online. I don't want to have re-rip a couple of thousand disks (though they're there I suppose). Brainy science says that most humans generally can't tell the difference between a reasonable bit-rate lossy compression and lossless. And as my wife points out, I mostly listen to music that sounds like angry women locked in an empty water tank anyway with nothing other than power tools for company.

[Little Jim]

Don't forget backups, otherwise that might be a lot of disks to re-rip...

I was going to use a 2 bay NAS drive with one mirroring the other to (hopefully) save me the faff of getting the CDs back out of the loft in the future.

[Kim]

Don't forget backups, otherwise that might be a lot of disks to re-rip...

I was going to use a 2 bay NAS drive with one mirroring the other to (hopefully) save me the faff of getting the CDs back out of the loft in the future.

If that's RAID1 mirroring, that isn't really a backup.  It allows quick recovery from most^[1] disk failures.  It doesn't protect you from the more common problem of accidentally deleting/overwriting the data, or extreme cases like the NAS going up in smoke or being stolen, data corrupted by malware or whatever.  If you've got the CDs, that's probably fine, but you might want to be a bit more paranoid about eg. photos.


_{[1] Every now and then, RAID will faithfully duplicate data corruption across the redundant disks.}

[Little Jim]

Don't forget backups, otherwise that might be a lot of disks to re-rip...

I was going to use a 2 bay NAS drive with one mirroring the other to (hopefully) save me the faff of getting the CDs back out of the loft in the future.

If that's RAID1 mirroring, that isn't really a backup.  It allows quick recovery from most^[1] disk failures.  It doesn't protect you from the more common problem of accidentally deleting/overwriting the data, or extreme cases like the NAS going up in smoke or being stolen, data corrupted by malware or whatever.  If you've got the CDs, that's probably fine, but you might want to be a bit more paranoid about eg. photos.


_{[1] Every now and then, RAID will faithfully duplicate data corruption across the redundant disks.}

Thank you, and yes, I understand what you mean, even though I have no idea what the difference between RAID 1 and 2 is (even if there is such a thing) but as I will still have all the CDs there is little risk apart from having to re-rip everything, and the photos have to be more secure than just sitting on the PC as they do at present.

[Kim]

RAID2 is one of the obscure levels that never gets used in the real world.

You're most likely to encounter:

RAID0 - Stripes blocks across n disks.  Space is the total of the capacity of all the disks.  Speed is improved significantly.  Failure rate multiplies (as any single disk failure causes you to lose all the data).

RAID1 - Mirrors blocks across n disks.  Space is the capacity of the smallest disk.  Speed is faster for reads, slower for writes.  As long as one disk still works, you keep all your data.

RAID4/5/6 - Stripes blocks across n disks with different strategies for creating parity blocks.  Space is somewhat less than the total of all the disks (how much varies with the level).  Speed varies, but generally slower for writes than the simpler levels, as parity has to be calculated and written (possibly more than once).  Level 4 and 5 can tolerate one disk failure, level 6 can tolerate two.

https://en.wikipedia.org/wiki/Standard_RAID_levels

If you're asking these questions, you're almost certainly considering hardware that gives you a straight choice between RAID0 and RAID1.  Avoid RAID0 like the plague, as unless you have a specialist application that needs very high throughput, it's just a way to lose data.

[Little Jim]

RAID2 is one of the obscure levels that never gets used in the real world.

You're most likely to encounter:

RAID0 - Stripes blocks across n disks.  Space is the total of the capacity of all the disks.  Speed is improved significantly.  Failure rate multiplies (as any single disk failure causes you to lose all the data).

RAID1 - Mirrors blocks across n disks.  Space is the capacity of the smallest disk.  Speed is faster for reads, slower for writes.  As long as one disk still works, you keep all your data.

RAID4/5/6 - Stripes blocks across n disks with different strategies for creating parity blocks.  Space is somewhat less than the total of all the disks (how much varies with the level).  Speed varies, but generally slower for writes than the simpler levels, as parity has to be calculated and written (possibly more than once).  Level 4 and 5 can tolerate one disk failure, level 6 can tolerate two.

https://en.wikipedia.org/wiki/Standard_RAID_levels

If you're asking these questions, you're almost certainly considering hardware that gives you a straight choice between RAID0 and RAID1.  Avoid RAID0 like the plague, as unless you have a specialist application that needs very high throughput, it's just a way to lose data.

  Brilliant, thank you! 

I'm now going to memorise all that and quote it at my computer bod when he comes to set it up for me!  (By his own admission he isn't a NAS specialist but he does seem to know the right people to ask to solve most problems)

[Woofage]

If you can allocate the budget I recommend adding an external (USB) backup disc of the same capacity (or greater) than the main disks. The NAS will be able to run a nightly backup to this disc, thus avoiding potential "oh shit" moments (if you want to store more than your music there, it's a must IMO).

If you've got your photos on your laptop, keep them there and do regular backups of these to your NAS.

[freeflow]

Repurpose an old pc or a top end raspberry pi and use XigmaNas ( formerly Nas4free) with a zfs disk array. If you value your data you'll set up two, one in the house and one somewhere else, and then use syncthing to keep them synchronised.