Most affordable way to securely store 5+ TB?

[doodler85]

Hi guys,

My friend is a photographer and needs to securely store at least 5 tb of data. He's not stuffed with money so I'm trying to think of a good way to store that (with backup) in the most reasonable budget possible. This machine doesn't have to always be on but I'm trying to think of something that won't require swapping out drives all the time.

The RAID enclosures I've looked at aren't cheap... they seem almost pricey enough to just build a home server ourselves. Would you trust building a cheap RAID-enabled PC? What do you guys think is the best solution?

Thanks!

 

[CurseTheSky]

5 TB is stepping into the realm where there really isn't a cheap solution, unfortunately. Even with a good RAID 01 / 10 (I always forget which is which ), 5, or 6, you're going to need some form of backup just in case. Lightning strikes do happen (even with surge suppressors), fires, floods, theft, etc.

The very first thing I'll say is try to avoid the current crop of 1.5 and 2 TB disks. They seem far more prone to failure than 1 TB disks with fewer platters. While the price / storage is very handy, it's generally not worth the hassle or cost of replacing a disk, especially if two decide to fail at the same time on you.

 

[Computers_Plus]

for about $370 you can build a 4 gig external storage


http://www.newegg.com/Product/Produc...-001-_-Product

and a couple of 2TB drives

http://www.newegg.com/Product/Produc...82E16822145369

 

[doodler85]

Hey thanks guys.

That MediaSonic box looks really nice! I've never seen that before but it's very tempting.

Question about that -- he has a Macbook Pro Laptop... if he were to connect via ESATA to a box like that would he have to worry about mobo-related space limitations? I have to do some research to see if it supports port multipliers as well... but this would be a nice start.

Thanks!

 

[Computers_Plus]

Not sure I understand your question about space ??

If he uses it via USB then you don't have to worry about the port multipliers either.

 

[RebateMonger]

There's two different needs here:

1) Provide 5 TB of convenient, reliable, storage

2) Back up 5 TB of data.

If data is important, you want at least two copies on two independent storage systems.

Plus, you need to decide whether the copies can be kept in the same location as the originals or if at least one copy needs to be kept elsewhere. Homes and offices are subject to fire, flood, theft, and other mishaps. Home burglary rates are higher than you might think.

 

[Computers_Plus]

Problem is there is no really cheap way to achieve all that at least with the solution I provided it gives you 4 gigs of storage plus you still have an open bay to add another TB or two storage later.

I think the cheap way out at least until you can afford a little more cash would be to just burn DVD's for the backups.

 

[doodler85]

I like the idea of that box... it's not as elegant, as, say a server that will handle all the RAID stuff for you, but you could at least use it as a central location for a ton of storage and do redundant back-ups manually or on a schedule to other disks / devices. I realize he could lose things in a fire or burglary but I don't imagine thieves would go straight for the hard drive array. So that's a risk we're ok with taking.

Back to the ESATA thing -- are all the hardware issues covered by the box itself (like USB does) or do you rely on the attached computer's motherboard? I was looking at some Apple spec sheets and they were talking about total storage limits on their hardware... so my question is, is that something that you need to take into consideration with a box like this?

Thanks again!

 

[tweakboy]

Sheez your going to need a lot of hard drives to store that stuff and I recommend external. WHY ? because you can turn it off and the 5TB is safe.

Just get 5 ,,, 1TB external usb 2.0 so its cheap price,, I would guesstimate 70 bucks a pop around.... soo ya that is my opinion on this matter for your friends photos... gl and gb

You dont want backup stuff to be on a mechanical drive my friend.. pointless one breaks down they all go , plus pc is gonna be on and those drives are gonna be spinning and if they get too hot with bad cooling some die on him.

USB 2.0 is the cheapest route its slow but its cheap. 5 1TB external usb 2.0 and then turn them off and hes got it all there in each drive which wont be used much soo they will never die on him. gl

 

[taltamir]

Hi guys,

My friend is a photographer and needs to securely store at least 5 tb of data. He's not stuffed with money so I'm trying to think of a good way to store that (with backup) in the most reasonable budget possible. This machine doesn't have to always be on but I'm trying to think of something that won't require swapping out drives all the time.

The RAID enclosures I've looked at aren't cheap... they seem almost pricey enough to just build a home server ourselves. Would you trust building a cheap RAID-enabled PC? What do you guys think is the best solution?

Thanks!

Securely as in... don't lose it due to a HDD failure, or securely as in "data requires security clearance and must be encrypted and unrecoverable if I want it destroyed"?
I am going to assume you meant "don't lose the data in case of drive failure"

Unfortunately 5TB is a whole lot of space... Since he is a photographer these must be individual photos, aka, many small-ish files (I imagine they are very high resolution)...

1. He should consider some good looking lossy compression... could save a LOT of space.
2. Two 5TB RAID NAS boxes built with something like freeNAS or freeBSD (for ZFS), and backing up the data from one to the other.
I recommend multiple SEPARATE arrays of RAID1 in each box. Then when he upgrades in the future, he can simply replace the smallest one as needed.

Sheez your going to need a lot of hard drives to store that stuff and I recommend external. WHY ? because you can turn it off and the 5TB is safe.

Just get 5 ,,, 1TB external usb 2.0 so its cheap price,, I would guesstimate 70 bucks a pop around.... soo ya that is my opinion on this matter for your friends photos... gl and gb

external drives:
1. EXTREMELY UNRELIABLE!
2. More expensive!

 

[doodler85]

Secure as in "I don't want my life's work to disappear if the drive goes out"

 

[taltamir]

Secure as in "I don't want my life's work to disappear if the drive goes out"

ok, works for me...
also, can you verify he really needs 5 TB and not 5 GB? 5TB of photos is a LOT of photos.

looking at newegg a 1TB drive is 60$, 1.5TB is 80$, and 2TB is 100$.
This makes 2TB the cheapest per GB. He can get 2x2TB and 1x1TB for a total of 5TB... but then you can't add anything more, so I will assume he is getting 6TB worth of space.

The cheapest way to do it is to put 3x 2TB drives in his current computer (no RAID)
3x 2TB drives in his wife's computer (no RAID)
Put all the data on his computer as a "main"
Share those drives over network (windows simple file sharing), and use a program like second copy (20$ lifetime license) to automatically backup from his computer to hers (scan for changes and modify only as needed)
This requires that his and her mobo and case be able to accomodate that many drives (they only take 5 watt on load and 2 watt idle... so no need to upgrade PSU)

that will be a mere 6 drives of 2TB... so a grand total of 600$... keep in mind 5TB is a LOT of data.
The best way is two seperate file servers running 3 raid1 arrays... which means 900$ PER server.

Mmmm... If I was doing it for MYSELF I would go with a single ZFS fileserver (ZFS is the ONLY currently available filesystem that protects against file corruption, extra drives do NOT, individual photos WILL be lost to random file corruption without ZFS), so that is 900$ worth of drives (9 drives) right there + cost of server, if he can cobble it from old parts (I did for my fileserver) then great, if not, he would have to spend a little, say, 200$ to build one...
and put 3 individual drives in his and his wife's computer (no raid) to serve as backup in case of something catastrophic (so that is 3 out of the 9 drives).

Ok, let me rewrite this in a less confusing and more clear and concise recommendation:

Option 1:
300$ 3 x 2TB Drives, not raided, in his computer (main storage)
300$ 3 x 2TB drives, not raided, in wife computer (backup)
20$ second copy software
total: 620$

option 2:
0 to 200$ A file server
600$ 3 x 2 RAID1 arrays of 2x2TB drives each, ZFS via FreeBSD (main storage)
300$ 3 x 2TB drives, not raided, divided between his and his wife's computer (the more computers it is divided amongst, the better), (backup)
20$ second copy software
total 920 to 1120$

I recommend option 2

Q: Why not use RAID5?
A: RAID5 is not only unreliable and slow, it screws you when it comes to upgrading... instead of replacing only the smallest drives, you must build an entirely new server with a new RAID5 array, copy things over, then somehow sell the old server and all the old drives at once. And you will only end up using 7 instead of 9 drives for option 2 with RAID5. I have a RAID6 server right now and I am totally stuck in terms of upgrading capacity... so I am just avoiding increasing storage and waiting for bigger and cheapter drives to switch to multiple raid1 arrays system (its got 5x750GB drives)

PS. Read up on different raid levels http://en.wikipedia.org/wiki/RAID

 

[doodler85]

Haha, yeah of course it's 5 TB You can easily shoot 5 GB in a single session, if not 10 or 20.

Could you give me some more info about ZFS?

Thanks again for all of your responses guys, this has been very informative!

 

[doodler85]

I'm rather concerned/ curious about what CurseTheSky said about 2 TB drives being more failure-prone. Anyone have numbers on this? % increase in fails, % difference in lifespan, etc.?

 

[taltamir]

I'm rather concerned/ curious about what CurseTheSky said about 2 TB drives being more failure-prone. Anyone have numbers on this? % increase in fails, % difference in lifespan, etc.?

1. With a proper backup system it doesn't matter.
2. Without a proper backup its going to be poor consolation that it was "less" likely to fail when it DOES fail and take your life's work with it...
3. It takes years for proper failure data to accumulate, those vary greatly on a per drive MODEL basis... all companies occasionally released lemon models. This is part of why it is so important to have a good backup scheme in place.

Haha, yeah of course it's 5 TB You can easily shoot 5 GB in a single session, if not 10 or 20.

Heh, I am not a photographer so I wouldn't know, but thanks for clarifying.

Could you give me some more info about ZFS?

Thanks again for all of your responses guys, this has been very informative!

ZFS is a filesystem developed by Sun Microsystems, it is open sourced and has been integrated into FreeBSD as well.
It kicks to the curb 20 year old assumptions that still exist in every other filesystem. It is fully checksummed so it will
1. catch data corruption
2. automatically correct said corruption if possible (aka, if you have any form of redundant raid)

Data corruption can come from many sources, power fluctuations (due to faulty PSU or bad wiring in your house or whatever), solar radiation randomly flipping bits, bit rot, microparticles scratching your drive, etc etc.
My own ZFS has cought and corrected 3 single bit data corruption events the last year which would have otherwise resulted in corrupt data. Those were probably due to solar radiation due to their rarity, I have a very high quality PSU & UPS ensuring clean power and everything else is running spiffily... people with faulty PSUs or other serious problems have reported IIRC thousands of errors cought and corrtected by ZFS... and it would have gone completely unnoticed without it.

you can read more here http://en.wikipedia.org/wiki/Zfs

My file server is running opensolaris. however, with oracle's acquisition of sun several months ago development of open solaris effectively halted (open solaris was managed by sun).
You can either use the latest open solaris beta (its pretty good and what I currently use) or switch to the new community development that picks up where open solaris dropped off (which is heavily supported by several other companies that have come to depend on solaris, but is community controlled, not corprate controlled... and uses the same code as opensolaris did, with a few minor changes), called Project Illumos. I am waiting a little bit before making the switch for illumos. Once my current osol version is a year old I will test illumos and FreeBSD and switch to one or the other.

ZFS though is OS independent. I have, just to test it out, formatted my fileserver's OS drive (an old 160GB drive ) from osol to FreeBSD, and imported the ZFS array (a simple single line command, done instantly) and tested the data was all there and accessible, then I did the same to go back to osol... and I also upgraded osol that way a few times, just install any OS that supports ZFS and type in:
ZFS import -f arrayname

 

[taltamir]

delete

 

[Computers_Plus]

The setup I suggested would work for your needs without breaking the bank yes there are other options but your cost is going to go way up from what I am suggesting.

Other posters remember he said we are dealing with a Macbook here not a desktop we can load full of drives.

As for drive failure that is always possible regardless of the type of drive or where it is stored what I would do is since it is an easy USB plug and play setup is plug in the backup array run your backup and unplug the drives that way you keep run time down to a minimum thus reducing your failure risk.

 

[taltamir]

The setup I suggested would work for your needs without breaking the bank yes there are other options but your cost is going to go way up from what I am suggesting.

Other posters remember he said we are dealing with a Macbook here not a desktop we can load full of drives.

As for drive failure that is always possible regardless of the type of drive or where it is stored what I would do is since it is an easy USB plug and play setup is plug in the backup array run your backup and unplug the drives that way you keep run time down to a minimum thus reducing your failure risk.

the setup you suggested is to store it on 5x 1TB external drives, with either no backup, or manual (painful!) backup to and fro to an additional 5x1TB external drives.
@70$ per 1TB external drive, thats a total of 700$, which is actually MORE expensive than my option 1 (for less space), and much much less reliable, and much more work for him.
He is better off with 2TB external drives, since 2TB Drives cost less per GB.

I didn't realize it was a macbook, that is a problem... OP, how many actual desktops does your household own?

 

[doodler85]

Not sure how I ended up in a household with a wife and all but I'll play the role if necessary

My friend has one 2005 Apple G5 Tower (2 HDD slots total, one being taken by his OS HDD) and his MacBook pro. I have two spare PC towers I can give him (one is a P4 2.8 ghz and another is an Athlon 3000+), so we're looking at some MacGyver solution I imagine. I would need to check how many SATA slots are open in each before I could tell you anything (and I'm not sure how HDD space limitation works with mobo's... is it really an issue?).

I'd love to build a server box with a million drives in it but I don't think he has the money for that.

 

[Computers_Plus]

You better go back and reread my post I suggested 2 2TB drives because it was the cheapest cost to storage ratio so he starts out with 4 gigs plus the array would still have an open bay so he could add 1 ot 2 more TB's to up capacity.

As for manual backup it's really not that bad start a DVD burn and walk away he may not need daily backups anyway maybe weekly would be fine on that yeah it's not the most awesome way mine is all automated but again he is looking at keeping cost down in order to do automated he would need to spend more money on hardware.

 

[taltamir]

Not sure how I ended up in a household with a wife and all but I'll play the role if necessary

I thought this was a project for your friend who has a wife, not for you personally... I must have mixed you up with someone from another thread who is asking advice for something else

My friend has one 2005 Apple G5 Tower (2 HDD slots total, one being taken by his OS HDD) and his MacBook pro. I have two spare PC towers I can give him (one is a P4 2.8 ghz and another is an Athlon 3000+), so we're looking at some MacGyver solution I imagine. I would need to check how many SATA slots are open in each before I could tell you anything (and I'm not sure how HDD space limitation works with mobo's... is it really an issue?).

Ok, your two towers are perfect towers to use, and with an awesome name too (since I don't see any way to fit them in his laptop or G5 tower).
He wouldn't need to be running windows on them anyways.
Get the twin towers you have, put drives in singles in each (no RAID), and set a backup scheme.

The McGuyvering is minimal, either they have enough SATA slots, or they can take a PCI or PCIe card that gives them enough SATA slots... since cost is of such importance we simply put 3x2TB drives in each tower. Total storage capacity is 6TB (with 6TB of backup in second tower).
Name them after the lord of the rings twin towers for extra coolness points.

You better go back and reread my post I suggested 2 2TB drives because it was the cheapest cost to storage ratio so he starts out with 4 gigs plus the array would still have an open bay so he could add 1 ot 2 more TB's to up capacity.

As for manual backup it's really not that bad start a DVD burn and walk away he may not need daily backups anyway maybe weekly would be fine on that yeah it's not the most awesome way mine is all automated but again he is looking at keeping cost down in order to do automated he would need to spend more money on hardware.

sorry, I confused you with tweakboy.
Your solution sounds... well, exactly like mine
Except for the DVDs... seriously... 5TB backuped to DVDs? thats over 1000 DVDs! and by hand? and besides, home grade writable media is CRAP in terms of reliability... of my 600 burned CD collection 1/3 was unreadable after 3 years. Factory pressed ones are only rated for 10 years worth of storage before the onset of bit rot (physical degradation of the chemicals from which the media is made)

Plus, you need to decide whether the copies can be kept in the same location as the originals or if at least one copy needs to be kept elsewhere. Homes and offices are subject to fire, flood, theft, and other mishaps. Home burglary rates are higher than you might think.

Urgh.... After all the effort I went to write a detailed data protection guide I went and ignored my own instructions... I am very tired, its the only excuse... if you look at my own guide here:
http://forums.anandtech.com/showthread.php?t=218081&highlight=how+protect+data

Basically you should protect against data loss causes, not just randomly throw about raids and backup solutions...

Maybe he should just use an online backup, been hearing ads for one called carbonite on the radio.. or just google for it. Mmm... it seemed too good to be true to get "unlimited backup" so I did a quick google on that specifically, turns out that carboniate gives you "unlimited" with the caveat that "abuse" will result in termination... what is abuse? over 125GB... so apparently "unlimited" is just a marketing term they use to lure in customers who have no idea of what a GB is.

a little copy pasta

People often implement some sort of redundancy or backup and think their data is safe. That is incorrect.

To correctly protect your data you must list all the possible ways in which data can be lost, and then address each and every one of them (or the ones you think are pertinent)

Here are a list of common disasters and the ways to protect against them:
1. Fire - offsite backup, DO NOT use a firesafe as they are meant for paper documents (350 internal temp max... which will ruin your data, but not your paper documents)
2. Flood - offsite backup, waterproof media?
3. Swatting: Swatting comes from the term SWAT, SWATting is when someone places an anonymous call to the authorities, and lies, claiming that you traffic in illegal materials such as illegal gun trading or CP.
Even if you are completely innocent, your computers and drives will be held as evidence for years afterwards. Offsite backup is the only protection against this (as it will surely cause your business to fail otherwise). Note that this isn't advice to circumvent the law, if you are guilty they will likely go after your offsite backups as well. But if you are innocent this can save you. (I have read stories of people who run small software business from home having that happen to them, there was never any evidence found of any wrongdoing on their part, but the FBI kept their computers and data)
4. bit rot - use a checksumming filesystem (only ZFS at the moment); create par2 recovery records.
5. Cosmic rays bit flipping - same as bit rot
6. Drive failure - use redundancy (raid 1,5,6, 10/1+0/01/0+1 etc); or backup. Note that neither RAID 0 nor JBOD are redundant and will not help.
7. Controller failure - use a scheme that allows you to replace the controller (aka, NOT a motherboard raid controller... motherboard RAID1 is ok though, raid1 is fully portable even on mobo controllers).
8. A controller failure with a controller no longer available on the market - use a system that allows you to switch to other controllers (aka, OS based software raid like in linux, solaris, windows server, or open nas; or use controllers from a company that historically makes newer versions support older controller made arrays).
9. Theft - offsite backup OR/AND worthless media backups (DVD-R). they will take your fileserver, iphone, external drive, and flash drives. but unlikely to take your DVD-R collection.
10. Virus - NOD32 antivirus , but ideally, have an external backup that you turn off / unplug except for when you manually connect it / turn it on to perform a backup. (not that i have seen a virus that just deletes data).
11. Drive erroring (due to a bad drive or PSU) but still writing - use a checksumming filesystem (only ZFS at the moment) AND redundancy (RAID1, 5, 6, etc). using only RAID will leave you with corrupt data as it has no way of knowing which copy of readable data is corrupt and which is correct. Using only checksumming filesystem will let you know it is corrupt, but without a non corrupt copy to recover from.
12. Sabotage - offsite rolling backup; internal security.
13. Lightening strike - backup to optical media (CD/DVD/Bluray), or backup to an external drive that you physically unplug from both the power AND the PC when not in use AND doesn't sit on a metal surface. If it is sitting on the PC case, or if it is plugged into the PC via a USB cord, that is enough to fry it!
Also, use surge protectors on your PC / Server / External drive (not as safe as unplugging it, but it might save your data and hardware)



Also, please note that whatever method of data protection you use, it is CRITICAL that you:
1. Check to see that the backups are happennig as expected. (is it really backing up daily? or did a careless employee disable it 9 months ago and did not turn it back on later? is the external HDD it is backing up to even powered on? ps. both are real stories i dealt with)
2. Check that the backups are readable.
3. Check that all needed files are being backed up (for example, your firefox bookmarks might not be backed up because you can't copy them while FF is running, as they are considered in use)



Terms explained:
Rolling Backup: Rolling backup is what you have multiple backups from different time periods. For example, you have a complete backup of all your files from 1 week ago, 1 month ago, 2 months ago, 3 months ago, and 1 year ago.
PAR2: http://en.wikipedia.org/wiki/Par2
ZFS: http://en.wikipedia.org/wiki/ZFS

 

[doodler85]

You better go back and reread my post I suggested 2 2TB drives because it was the cheapest cost to storage ratio so he starts out with 4 gigs plus the array would still have an open bay so he could add 1 ot 2 more TB's to up capacity

Regarding that... does his laptop's mobo affect how much memory it can handle through an ESATA expansion card? Or is it handled like most external drives handle USB (where it's irrelevant)? Sorry for asking this so many times, I just haven't got a clear answer yet.

 

[Computers_Plus]

doodler85 if he has a G5 you can add more drives to that or take one of the towers you have and make a file server with that

 

[taltamir]

Ok... I looked at some more honest online backup schemes and I am seeing things like 2GB/$ per month... at 5TB we are talking about 2500$ a month... this is utterly ridiculous...
Since offsite backup is the only way to protect against theft, well... find a friend or a family member willing to hold onto a fileserver for you.

 

[doodler85]

I'm seeing on the 2005 G5 documentation that you can only have 2 internal drives... are there other expansion options?