this post was submitted on 30 Nov 2023
3 points (80.0% liked)

Data Hoarder

116 readers
1 users here now

We are digital librarians. Among us are represented the various reasons to keep data -- legal requirements, competitive requirements, uncertainty of permanence of cloud services, distaste for transmitting your data externally (e.g. government or corporate espionage), cultural and familial archivists, internet collapse preppers, and people who do it themselves so they're sure it's done right. Everyone has their reasons for curating the data they have decided to keep (either forever or For A Damn Long Time (tm) ). Along the way we have sought out like-minded individuals to exchange strategies, war stories, and cautionary tales of failures.

founded 1 year ago
MODERATORS
 

So I run a video production company. We have 300TB of archived projects (and growing daily).

Many years ago, our old solution for archiving was simply to dump old projects off onto an external drive, duplicate that, and have one drive at the office, one offsite elsewhere. This was ok, but not ideal. Relatively expensive per TB, and just a shit ton of physical drives.

A few years ago, we had an unlimited Google Drive and 1000/1000 fibre internet. So we moved to a system where we would drop a project onto an external drive, keep that offsite, and have a duplicate of it uploaded to Google Drive. This worked ok until we reached a hidden file number limit on Google Drive. Then they removed the unlimited sizing of Google Drive accounts completely. So that was a dead end.

So then we moved that system to Dropbox a couple of years ago, as they were offering an unlimited account. This was the perfect situation. Dropbox was feature rich, fast, integrated beautifully into finder/explorer and just a great solution all round. It meant it was easy to give clients access to old data directly if they needed, etc. Anyway, as you all know, that gravy train has come to an end recently, and we now have 12 months grace with out storage on there before we have to have this sorted back to another sytem.

Our options seem to be:

  • Go back to our old system of duplicated external drives, with one living offsite. We'd need ~$7500AUD worth of new drives to duplicate what we currently have.
  • Buy a couple of LTO-9 tape drives (2 offices in different cities) and keep one copy on an external drive and one copy on a tape archive. This would be ~$20000AUD of hardware upfront + media costs of ~$2000AUD (assuming we'd get maybe 30TB per tape on the 18TB raw LTO 9 tapes). So more expensive upfront but would maybe pay off eventually?
  • Build a linustechtips style beast of a NAS. Raw drive cost would be similar to the external drives, but would have the advantage of being accessible remotely. Would then need to spend $5000-10000AUD on the actual hardware on top of the drives. Also have the problem of ever growing storage needs. This solution we could potentially not duplicate the data to external drives though and live with RAID as only form of redundancy...
  • Another clour storage service? Anything fast and decent enough that comes at a reasonable cost?

Any advice here would be appreciated!

top 50 comments
sorted by: hot top controversial new old
[–] [email protected] 3 points 11 months ago

NAS.

Over the last 24 months I've built 300TB (a mix of 10 and 14TB disks) for $2500 in disks. I could do that right now for $2100. A 18TB LTO9 tape is more expensive than what I'm paying per TB for 14TB disks.

$700 in hardware to build the NAS with 25 bays.

Glacier would cost you $1080/mo in storage fees alone (300,000GB @ $0.0036) not including the $0.09/GB to get any data back out. Deep Glacier is less (by half, for storage), but comes with strings attached.

Don't forget to factor in labor hours of what it's going to cost you to maintain a tape library or a local server in general.

Are you charging clients for long term storage after a project is complete? If not, you should be.

[–] [email protected] 3 points 11 months ago

AWS Glacier Deep Freeze is designed for this. Something you access a couple of times per year if that, and it's $.99/TB/mo. Price that out compared to a $10k NAS or tape backup that will still need consumables like drives and tapes, and it might be your best option. There are costs on retrieval, but since as you've said this is archive footage that customers might request you could pass that cost down to them.

[–] [email protected] 2 points 11 months ago (3 children)

Have you considered Amazon S3? It’s made for enterprises with unlimited storage, a lot of pricing options and could save you a lot of headaches long term.

[–] [email protected] 2 points 11 months ago

Backblaze B2 is cheaper by a long way.

[–] [email protected] 1 points 11 months ago

s3 is designed with high availability and high throughput in mind, op needs a cold storage solution like aws glacier or azure cold storage. but even that is not cheap

[–] [email protected] 1 points 11 months ago

Yeah we looked into it. But as subven1 pointed out, it's a brutal monthly cost.

[–] [email protected] 2 points 11 months ago

You have 3 issues, online archive of past projects, long term (offline) storage & client access.

LTO is your long term solution for offline archive of projects. Depending on the average / largest project you might want to do 1 project per tape so LTO7/8 sizes. Scales really well, multiple copies, etc.

For the online storage, a NAS is really the only option. How it's sized & configured comes into play. You can go cheaper with used enterprise gear, but then you're dealing with more disks & higher power bills. Fewer larger disks can help with the power bill & noise levels.

Splitting things between a read-only share (of things that have been archived to tape), and a normal working share would help on the workflow.

The catch is what you do for client data exchanges. Giving them access via Dropbox is nice, but you need better housekeeping around data. Once the 1 year grace is over, what's the size they have committed to? While self-hosting a client accessible share is possible, there's ongoing costs & I would be cautious around exposing the NAS to the internet directly.

[–] [email protected] 1 points 11 months ago (1 children)

https://youtu.be/lO-SAzFaN18?si=Rp0mvidHMxBFNedC

https://youtu.be/JHVSoJDZ06U?si=_7kmEUZNKc3UDfRK

I'm by no means qualified to give you an actual answer, but these videos may be of some use to you.

[–] [email protected] 1 points 11 months ago

Yeah sweet. I haven't checked in on the Slow Mo Guys storage setups in a while. I'll have a watch.

[–] [email protected] 1 points 11 months ago (1 children)

Really depends on how often you need to touch your data. Tape has high upfront cost (4-5k $ for a LTO-9 tape drive + ~3,5 $/TB in tapes) but you don't have to worry about archive space anymore. Otherwise, NAS space (if you selfhost) is ~15 $/TB + a server which would also be slightly above 5k right now to store your 300 TB.

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago (2 children)

How often are you actually needing to access the 300TB? If 250TB are “cold storage”, then LTO is the way (you can rent the readers usually, rather than buy)

If you’re needing to have access but not edit from, NAS is the way, 300TB wouldn’t even be THAT expensive (still expensive), just slow to move to, but once you’re up and running a decent rig should last years.

If you’re needing to access all 300TB, then you’re looking at a LTT style NAS that needs to handle read and write from multiple users at a time, and that’s gonna be the real $$$.

I feel like you might do well from a mixture of all of these. A smallish NAS for day to day/project use, and once that project is done you move it to the big “slow” server for onsite backup, and once every 2-3mo you rent the LTO drive and load up a few tapes, and ship them off to the void for offsite backup and cold storage.

[–] [email protected] 1 points 11 months ago (3 children)

Just copying from a response above:

This is only for archived projects. But we'd probably still need to access ~10-20TB of that data relatively regualry to update branding, or change edits, etc. Saying that, as mentioned in the OP, if we went tape or cloud, we'll likely have a physical local copy on an external hard drive for quicker access. We just need a redunant back up of these archives.

If we went NAS, I feel like maybe we could get away without the redundancy? Risky...

[–] [email protected] 1 points 11 months ago

If we went NAS, I feel like maybe we could get away without the redundancy? Risky...

That's the thing, you could, but it wouldn't be best practice. At the end of the day the 3-2-1 rule applies to any data.

I know it's a hard pill to swallow, but ideally you'd need both a NAS (I'd go with Proxmox on a PC) and the tape backup for that NAS to ensure the safety of the data.

However. Backblaze may take the spot of the tapes - unsure if the NAS as well. Have a look at their offer and see what fits your budget. I would personally go with the NAS on site and backup it daily to Backblaze. Note that Backblaze B2 says something like 6$/TB/month which amounts to about 21600$/year which stings but then again it's safe and it's the best value (all the competition seems to be more expensive).

[–] [email protected] 1 points 11 months ago (1 children)

Which 20TB? The last projects or random?

[–] [email protected] 1 points 11 months ago

More random than I’d want. Generally predictable though.

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago (1 children)

idk you can rent them.

oddly thank you for that info.

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago (4 children)

I don't know that I'd take on tape with your use case. There's a good bit of tech debt involved there.

NAS (either bought or built) + Amazon glacier or Backblaze for cloud archival backup.

The NAS (including drives) will probably cost you $7000-8000 USD for 400ish TB of storage with room to grow

It was easy to give clients access to old data directly if they needed, etc.

I hope you charge for this. It would help to offset your storage costs.

load more comments (4 replies)
[–] [email protected] 1 points 11 months ago (2 children)

From what it sounds you want a NAS and Tape Archive.

So get a device which holds your working Projects, you mentioned arount 20-40TB which is no problem nowdays. Can be done for under 1k with of the shelf stuff.

And Tape backup for stuff you dont need regularly. Maybe chose an older generation of LTO I would look for something that can hold about 1 Project per Tape or the likes of it. LTO5 is pretty cheap used, ca be had for 500 Bucks but is only 1.5TB per tape.

Disclaimer, with LTO never look at the compressed NR, its for compressable data only which video is not. Thus with LTO9 you will only get 18TB

[–] [email protected] 1 points 11 months ago (2 children)

Yeah we've got a solid situation for our live projects. Each of us work off 40TB thunderbolt raids with local external drives as our backup and live online backup to Dropbox.

This is for our archived work, but yeah of that, we access around 20-40TB fairly regualrly. Good to know that tape won't compress video data at all!

NAS is sounding more and more like our best bet.

[–] [email protected] 1 points 11 months ago (1 children)

Not to be rude or anything, but External RAIDs individual to the user is not really a solid soulution. It may work for 1-2 People working on one project at a time. But it just does not scale. What if someone needs to acces files of that project? they move the raid or plug their laptop on a differen workspace? Not really a great soulution IMO.

Like you say in the last part having a NAS with maybe a bit of room to grow sso 100TB might be the best option that way everyone can access the data and work accross projects. And more importantly it would offer work from a different place in the office or even work from home.

Yea with tape the compressed nr are very missleading. Thats a best case scenario where the files compress 2:1 with TAR+gzip which it literallly never does. Bestcase I have seen was 1.2:1 on a folder consisting of config files. Basically nothing nowdays is compressable you will interact with, except textfiles depending on format. So its best to always asume the raw space as the space you get

[–] [email protected] 1 points 11 months ago

Haha we’ve been this way for 12 years. Certainly not ideal if we scale. But we won’t ever. 4 of us ever needing access. And transferring over the network is not an issue. NAS is too slow for most real time editing. 10gbe is fine but still fairly slow. Those raids will soon be upgraded to SSD raids for each editor. Thanks tho…

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago

This.^

2 small NASs + 2 LTOs (LTO5 may be sufficient for your individual projects, but you also need to backup the NAS, so at least LTO 7 or 8)

[–] [email protected] 1 points 11 months ago

Did I read somewhere there are two work/office locations?

[–] [email protected] 1 points 11 months ago

I'm in the same biz. I use tape. Specifically a Mac mini + canister from guys that make Hedge. I then index each tape with neofinder, it makes it easy to find and pull projects. The idea was to make a system simple enough that it wasn't one persons full time job.

[–] [email protected] 1 points 11 months ago

This is a bit out there, but have you considered encrypting it & putting it in the Filecoin network? According to a network explorer, they've got ~24EiB of space online currently.

Apparently, alot of the time storing is done for free because the miners need valid data to store to qualify for network rewards.

In any case, DeStor has a contact form that it looks like ought to get you a quote.

[–] [email protected] 1 points 11 months ago

RAID is not a backup! A single raid array in a single server is still only one copy and one very big single point of failure.

[–] [email protected] 1 points 11 months ago

Back blaze B2, or Wasabi seem to be a lot cheaper than going AWS S3 I've checked with wasabi a while ago and there are no fees for downloading/uploading. $6/TB/month

[–] [email protected] 1 points 11 months ago

Maybe something like that can be a part of your solution? : https://www.hetzner.com/dedicated-rootserver/matrix-sx

[–] [email protected] 1 points 11 months ago

I'd recommend the hybrid approach with NAS and Tape Drive

Build a robust NAS system for remote accessibility, but consider setting up a hierarchical storage management (HSM) system. Frequently accessed or recent projects can reside on the NAS, while older and less accessed ones can be automatically moved to more cost-effective storage.

Invest in LTO-9 tape drives for archival purposes. While the upfront cost is more but tapes provide long-term, cost-effective storage. This is particularly useful for archival data that doesn't require frequent access. It adds an extra layer of redundancy and security.

[–] [email protected] 1 points 11 months ago

backblaze is probably the cheapest cloud option. But it might still be too expensive.

[–] [email protected] 1 points 11 months ago

Raid/NAS, as many others have said, isn't a backup.

However, you could have a single NAS and backup to AWS Glacier where storage costs for larger files is cheap going in and getting out in DR scenario is expensive, but maybe covered by your insurance depending on the DR event.

[–] [email protected] 1 points 11 months ago

I like doing this math.

A DS1821+ with (2) DX517 expansion bays would cost 4.1K AUD presuming 10% tax and would be 307 TB presuming (18) 22TB drives with a BTFRS file system running SHR-2 (allows for 2 drive failures).

(18) 22TB drives @ $22/tb AUD = $9.5K

So an all in cost for 307TB is 13.6K AUD using that equipment. 27.2K AUD to have a mirrored backup, but it sounds like you're ready for another 300+ TB right now, so 54.4K AUD to have 1:1 backups and 307TB of runway.

If AWS Glacier is what you're comparing to, then you make that up in 6 months.

Rack mount would be more convenient, as you can have 1PB volumes and a little less cumbersome and tidy setup - the 1821+ with expansion bays are 108TB max per volume, so you'd have to deal with 6 different volumes but maybe not a big deal if your filing system is by year/month. But getting into rack mount with Synology for example would basically double your infrastructure cost. Or you bite the big bullet now on scaleability and use a 60 Bay rack mount @ 29.9K AUD for just 1, but it's still roughly the same cost per drive bay as the 16 bay.

[–] [email protected] 1 points 11 months ago (1 children)

You should follow in the steps of Linus Tech Tips and build a massive file server. That's probably the most comparable system and situation to your case.

Ex

https://m.youtube.com/watch?v=pLC0FUnko-M

https://linustechtips.com/topic/1447503-this-is-my-endgame/

I'd then recommend tape backups that you store off-site , maybe backblaze , but as others stated that might be too expensive.

[–] [email protected] 2 points 11 months ago (1 children)

Please don’t model your IT systems after a YouTube channel that documents a catastrophic failure approximately once a month.

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago (1 children)

I work in a TV production company. Masters and rushes are archived to LTO8.

Drives are cheap but a real pain to keep around and you can’t keep them indefinitely.

But you’ll likely want a library. These are expensive. Not necessarily to buy, but to license and get tot software. I think our entire system, (library, 2xlto8 drives, server, software and licenses cost 15-20k)

And we only licensed 25 out of the 50slots. As it’s a real fucker as you have to license the slots twice, once on the library and again in the archive software.

But it’s been an absolute godsend, having archive projects available makes life so much easier.

[–] [email protected] 1 points 11 months ago

Love this. Do you know who you use for your library licence?

[–] [email protected] 1 points 11 months ago

I hope you charge your clients for archiving purpose. I work in a similar field as yours and no chance I'm archiving this much data if the clients aren't paying for this. I have a contract that stipulates assets are kept for 1 year then they pay a yearly fee for archiving or they agree that we may delete it.

[–] [email protected] 1 points 11 months ago

Have you looked at Wasabi or Backblaze? Possibly cheaper. It’s always cheaper to do this via a nas in the end. Big Synology or two smaller units at each site with expandability.

[–] [email protected] 1 points 11 months ago

Go back to our old system of duplicated external drives, with one living offsite.

That would be the cheapest method I suppose

assuming we'd get maybe 30TB per tape on the 18TB raw LTO 9 tapes

LOL. No. You're going to get 18TB per tape. If anything you're probably going to get less

Would then need to spend $5000-10000AUD on the actual hardware on top of the drives

You don't have to(?). You're only going to be servicing at 1Gbps at a time. Interface limitations aside, even a $30 raspberry pi can do that.

LTTs set up is different because they have high speed networking, multiple users requesting a ton of data all at the same time, and ZFS dedupe going on. You don't.

This solution we could potentially not duplicate the data to external drives though and live with RAID as only form of redundancy...

Not sure what this means

Another clour storage service?

S3 glacier is potentially a solution, provided you (or someone at work) is willing to put in the time and effort to read the fine print. I have a post here on the subreddit if you want a somewhat summarized version.

Even shorter version : 300TB = USD $300 per month. Every single terabyte you want to get back is $100 on top of that. Do the math on how much egress you need.

Oh, and again, don't forget to read the fine print. Because you WILL get screwed over if you try to use it as a drop in replacement for google drive.

[–] [email protected] 1 points 11 months ago (1 children)

Many comments here suggested AWS S3 Glacier. Just wanted to let you know that transferring this amount of data might be faster (and more expensive) with AWS Snowmobile.

load more comments (1 replies)
[–] [email protected] 1 points 11 months ago

Copy 1, long term storage: LTO / LTFS is the way to go for one of your copies, it is by far the most reliable storage solution.
Copy 2, a big ass Truenas build, they can be done at the cost of the hard drives and can be configured for very high availability with 3 parity drives. Pro tip, have spare drives in the chassis but not configured as hot spares, it's a long story, you will learn if you go down the path of truenas. Get a used chassis for the drives and hook it up with a SAS adapter.
Copy 3, if you have important data, you will need a 3rd copy, and that should be LTO as well.

If you want more details, feel free to msg me

[–] [email protected] 1 points 11 months ago

NAS plus 80 drive LTO9 Library with drive.

[–] [email protected] 1 points 11 months ago

Please keep us posted on what you decide. I am facing almost the exact same problem.

load more comments
view more: next ›