cross-posted from: https://lemm.ee/post/39437325
Sigh…
A couple of years ago there were discussions on how stupid 20+tb harddrives were, mainly because they are so slow that the time it takes for files to transfer to a spinning disk was too long.
Let’s say you have a good 20tb drive and it can transfer files at 200MB/s. To fill that drive, it’ll take 1 day and 8 hours of continuous transfer. If it’s failing, and you’re trying to get as much off of it you’re screwed.
Now let’s think about that micro SD card. It’s 4tb, and let’s be gracious and give it a v90 speed class. That’s 90MB/s. Looking at a calculation for the time it takes to fill it up, we’re sitting at about 14h and 14 minutes. Worst part is that SD cards don’t have SMART, meaning you don’t know when they’ll die.
From my experience, even good SD cards die in my raspberry pi running pihole, and the cards runs idle almost all the time.
Also there’s this thing that the higher capacity a storage device gets, the more valueable the data stored on it becomes, not directly because it’s high capacity, but because it’s more trusted by the user.
Guys, gals and anyone in between, please get a proper storage solution, something that won’t fail spontaneously. If you need that kind of capacity, go for a Nas with spare drives, or at least get an ssd.
/end rant
Not all use-cases require a high speed:capacity ratio.
I mean, I have an 18TB USB hard drive, which sustains transfer at about 50MB/sec in practice. It is nearly full, and its level of performance has never been a show-stopping problem.
It’s hard to imagine a use case where a NAS would be a viable alternative to an SD card.
I’ve had a usage tier for storage that looks like this
Temporary storage
- SD cards - unreliable storage you use temporarily to store pictures and videos before inevitably moving them to a more reliable and permanent solution.
- USB drives (hdd ssd etc) - used for when you you want to move files faster or more conveniently than over a Lan.
Permanent storage
- Nas, internal drives, tape drives, etc - for when you want to store a lot of data with configurations that allow you to use redundancy.
The issue with super high capacity SD cards for me is that they’re still fragile and prone to failure. When you allow someone to store that much data, it’ll be used as a more permanent medium, and since it has a lot of storage capacity you end up with a bigger data loss when it dies. Imo having 30 128gb SD cards would be better because if one dies or breaks, you lose 128gb and not 4tb.
Tldr I think 4tb micro sd cards are stupid.
I mean, it’s where I keep all of my important tax documents in pdf and my old family videos. It’s plugged in this here chromebook. Haven’t needed to take it out since I got the thing during a sale for $160. The chromebook that is. I don’t remember what 16Gb cost back then.
This is obviously not for large scale storage. But for stuff like cameras, which uses ever larger files for raw images
I totally get that… Here’s the thing though, at least in Norway a 1tb micro sd card costs 2200kr (~$203). If we extrapolate the price for a 4tb one, that’ll be 8800kr(~$813). If you or a company has the kind of money to spend almost a grand on a storage device, doesn’t that mean that the footage/photos are pretty valuable? If you had the kind of money/were going to record super valuable footage, wouldn’t you work hard to use cameras/recording systems that were capable of recording to redundant drives?
What I don’t get is what market section this product would even fit in. It’s too expensive for regular consumers, and also has terrible value. It’s not good enough for professional settings because it has no drive monitoring, nor does it have redundancy. It isn’t fast enough for the kind of footage that would require that kind of space(unless you’re recording a month long realtime video).
Also imagine how horrible the transfer speeds would be for individual photos when the os has to initiate a file transfer. If we say each photo is 20mb, that’s almost 200k photos. Yikes…
The raspberry pi is about the worst case scenario for SD cards. It may be idle, but an operating system is still making constant reads and writes, which absolutely eat through an SD card
There is a thing called Log2Ram that can help with it for Pis. I run it in my PiHoles
Let’s say you have a good 20tb drive and it can transfer files at 200MB/s. To fill that drive, it’ll take 1 day and 8 hours of continuous transfer. If it’s failing, and you’re trying to get as much off of it you’re screwed.
this is kind of why we have RAID, but arguably, you should literally just not be using RAID as a backup. Failing drives should be prepped for in advance, rather than dealt with in real time at the 20+TB scale.
The primary advantage to such dense HDDs is price, and power efficiency.
Also there’s this thing that the higher capacity a storage device gets, the more valueable the data stored on it becomes, not directly because it’s high capacity, but because it’s more trusted by the user.
also im not sure i agree with the phrasing here, the drive does become “more important” but that’s because it stores more data, there is literally more for you to lose in the event it gets destroyed. You should trust nothing ever, yourself included.
Worst part is that SD cards don’t have SMART, meaning you don’t know when they’ll die.
I mean, SMART doesn’t help much with knowing about HDDs’ death either. It’s more often they don’t show up at all, so you can’t even check SMART.
It’s not about the death. It’s about if it’s going to die. I’ve seen smart errors weeks before a hdd died which gave me time to back that data up.
Huh. I can see drones, action cameras and spy cameras being able to store lots of super high quality footage with this. Like, so much footage it lasts longer than the battery.
It’s niche, but I can see the use case.
What would anybody even use 4 TB SD card for? Storing a shit-ton of pirated movies that you can watch on your phone? Aside from that I have no idea. 256 gigs is probably more than enough for anything a normal user would do on a phone.
File size is a major limiting factor in high speed video and to a lesser extent convenient ultra HD digital film. At 3840x2160 (basic 4k) uncompressed 10-bit video 1 frame is about 250 MB. An hour of footage at 30 fps then is about half a terabyte. At “only” 1000 fps you would burn through an 8 TB SD card in… 32 seconds.
ah finally, i can buy a micro sd card for 500 dollars, the same price as a gazillion terabyte harddrive, and get less reliability out of it.
yeah, but you can carry it with you at all times if your phone takes an SD card.
although, can they use one that large, or is there some restriction?
this is true, my phone supports up to 400GB but it’s a bit older. Anything over about 512GB and you’re gonna run into issues writing and reading data reliably/fast enough. I’ve yet to find a way to transfer more than like 5Gb of files reliably to my android lmao.
It’s just a shit platform with shit software implementations, there’s not really much you can do about it.
I feel like this is a product looking for a market. Why would anyone ever trust that much data to something so fragile and easy to lose?
4k Drones, upgradeable phones, DLSR cameras, Data per weight etc.
I own a 1tb ssd for my Steam Deck, literally 0 complaints, runs real fast, can’t feel any heat, never need to take it out other than if I’m factory resetting, it’s perfect! (though Valves next deck should just have a bigger ssd slot)
Filming 8K in a raw format maybe? (a lot of cameras only have an SD card slot, or only the sd card slot is fast enough to record raw at higher resolutions)
You probably wouldn’t need to take it out of the camera either? so the form factor wouldn’t be major concern.
Also could be handy for smuggling banned movies/tv shows into authoritarian countries that block or outlaw unauthorised VPNs?
They’re not for long term storage, they’re for transient storage like photography, in particular stuff like surveillance cameras
if you want long term CCTV setup properly you should be using ethernet connected security cameras and then transmitting it back to a central server with a hdd always recording. It’s much more reliable and way more cost effective, just requires you running an ethernet cable to where the camera is.
It depends on the type of location, small remote locations might not even get their own local network
I’d love them for my dash cam if they were affordable. My camera records in front and behind of my van in 4k, so that’s 90-100 gigs an hour. I leave it running as a surveillance camera when I’m parked, so just going to work and back in one day would use over a terabyte.
Why would you continuously record? Just record motion events and you won’t need a card that large at all, plus it will last a lot longer.
