SD vs SSD what the difference and why one cannot fully replace the other?

since peoples cards keep breaking, clearly the reader wasn’t the issue :smiley: but the endurance of that card, if it was used for storage it would have been just fine… and what do you need 30-80gb at ssd prices that’s 3-10$ if you compare it to ssd prices…

so you are telling me you wouldn’t just use a smaller card which would actually last and run at peak possible speed of your hardware.

looks like the prices are surprisingly enough, about the same as ssd prices, ofc the more space you want to cramp into such a tiny card, the price goes up…
but if you want a very sensible 32-64gb it seems to be 10-15$…

so you want to spend a sata port… the thing you actually in storj / tardigrades case, the limited resource, to host your operating system on an ssd, which is much less compact, requires more power draw, and the sata port… just to boot…
how many do you get of those sata ports… two or four before you run out of options.
sounds more expensive to me, instead of buying an IMO inexpensive card that will last.

instead people buy a 5-6$ card or a 25$ card with a shit ton of space, burn it out in months or maybe a year, with low performance, and then the system crashes on them… their settings are lost, their storagenode might get damaged.

yeah 11$ for an endurance sdcard really seems like they are trying to squeeze the life out of people.
and totally not worth it :smiley:

it’s like walking around without shoes… but hey it’s cheap.

I would love to know where you can find NAND SD cards for so little money? There are some here and the prices are not even shown there. We had the topic a few months ago.

One should not underestimate how often read and write processes take place. I am thinking only of PiHole that logs all requests and generally all logs. There is a reason why the logs at Raspiblitz are outsourced to the HDD / SSD. Let’s assume that I am now a complete beginner and have, for example, InfluxDB on the SD card. An SD card won’t do that for long, you will notice after a very short time that the card is slowing down. I don’t want to say that the SD card will then break, but you will notice that an SD card is not made for such use. SDCards are designed for cameras, smartphones and such. But not as a root file system for Linux.

I can talk for a long time now. I would suggest you buy a Pi and cheap and expensive SD cards and see for yourself. You may then be able to have a say from your own experience and not just from google.


why should nand flash chips cost differently depending on what interface they are in, ofc there isn’t nearly as much size to work with so one will have to limit expectation a bit and ofc the bandwidth for the card will be much much lower due to fewer chips in the memory complex of the card.

so one won’t see any amazing feats, on top of that the manufacturers seem to be unwilling to disclose which memory cells they are using in the sd cards and have their own definitions on them…
this can ofc be detrimental in many ways, because most people won’t know what they are working with and will most likely put way to much data on a card thats not rated for it…

been running a boost disk on a sandisk ultra microsd for a few years now only running a really old laptop with way to little memory and which cooks any ssd i put into it, so i ended up with that solution… it’s been running quite well and is essentially a page / memory swap drive…

duno what’s inside it, and i ofc haven’t gone over 60% of its capacity to limit the wear pattern on the cells whatever they are, would a 500$ SLC PLP micro sd be better… sure… but like you said, then it would most likely be the interface / controller that becomes the limitation…

but why would anyone even consider using such a card for a RPI, that would just be weird.
putting a 500$ sd card in a 30$ computer, from my own limited testing with using a micro sd card it seems fine… but it was done correctly and it was a card rated for a certain endurance level / best performing on i had on hand and thought it would be a fun experiment.

i think Sandisk Ultra cards work fine for the job, or they do for boost drive in windows 10…
atleast the one i have.
granted the laptop haven’t seem immense usage, but it’s been used and the performance of it is greatly improved due to it running a spindle drive and has like 4gb ram, so i’m fairly confident it uses the card a ton when working.

duno if the card got slower, maybe… didn’t monitor it that closely
and the recommended sandisk extreme on the link i shared is the bigger version, but the same series and to my understanding the same cells on it.
where it got a 2nd place… a bit surprised they still sell them because i must have had this card for 6 or more years. does seem like the one on the list is a better version and mine isn’t a pro either…
but it’s seen a lot of use over the years.

so yeah… sorry to disappoint.

did a bit of digging, couldn’t verify that these are legit tho…
tried to find a verify the pricing from other vendors
but who knows maybe they are just an older no longer produced series, which is why they are more affordable…

Because of varying expected sales in different markets

You know, that is why SAS HDDs cost more than their SATA counterparts

true… i mean similar interfaces, SAS and SATA are very different technologies, granted…
but one can certainly get decent endurance on an affordable micro sd card, ofc the bigger the chip complex gets, such as in an ssd which has tons and tons of “chip” i mean cell’s obviously… use a bit more use to the SSD lingo.

an SSD would scatter the wear across many more cells and thus gain much higher endurances, due to it’s size, so SD cards would ofc always be interior to most if not all SSD’s
the basic rules of how it works and the endurance would or could be the same… ofc SSD NAND flash might be more difficult to put into the SD form factor, or have other problems with it… which is why the manufactures have unique processes for that particular task…

but the theory behind it will be much the same and be in part also affected by developments in the SSD NAND flash market / technology.

but not really that well versed in the SD card aspect of storage.
have been thinking of making the old laptop into a host to replace my outdated smart tv, i think ill try and do that with linux and then install it on the sandisk ultra micro sd card just to see what happens and how it runs.

one of the big problems i see with the SD cards is the capacity… because if they are using something like QLC or similar technology… then one runs into the same issue as on an SSD.
in the case of QLC because they are Quad level cells, one would essentially only have 40% of the capacity before significant slow down of the cell complex would start to happen, due to the caching in SLC format and then the transfer into QLC format later as the cell complex / controller chews it’s way through the cached data stored on QLC cells but running in SLC mode, thus demanding 4 bytes for every 1 byte stored, until transferred to QLC.

Thus the capacity usage on a SD card and SSD has very defined limits which will quickly created damaging wear patterns on the cell complex, before most users suspect…

this is further more not helped much by people having been very use to HDD’s which doesn’t wear to the same degree even if filled to near maximum capacity lets say 80% to put a number on it.

to do an example of a SD card / SSD memory cell capacity problem…
if we imagine a half filled QLC type flash … so 50% is free for caching, but thats QLC so if we transfer even about 10% of it’s full capacity at that point, we are overloading it.
because the 10% would take up 40% in SLC mode on QLC’s

ofc this is a static example and the reality would be a lot more fluid, but it’s very easy to overload modern flash memory cell complexes.

You can read this article to have better understanding a difference:


i liked my explanation much better as the one linked glossed over critical details important to the proper utilization of the SD cards, even if he went a bit more in depth with the controller aspect which i kinda glossed over.

really i don’t see why one wouldn’t have the more advanced parts of the controller inside the interface instead… doesn’t really make sense to put it on the storage media, expect for perhaps in case of mobile storage, there is also the option of letting software control the distribution of the data on the SD card, ofc that would require some sort of exchange of information between the interface and the card to perform that, about the cell / chip complex

the article barely even touches on any subject that i didn’t cover in details.
and he said more or less the same as i did, just not in so many words.

To run an operating system off a SD or MicroSD card it’s recommended to not go beyond 40-60% of max capacity, it’s best to format / partition the card with this restriction in mind…
or otherwise restrict space usage.

60% is the endurance and best capacity configuration
30-40% is the performance / endurance configuration

A quality brand endurance model SD card , by quality brands i mean such as Samsung or Sandisk and i’m sure many other leading/top memory manufactures.

Going beyond the 60% capacity mark on an high activity use of an SD card or similar compact flash cards will in most cases result in the card wearing itself out and the data may be lost.

The reasons for this is explain in this thread.
heat can ofc also destroy or damage the cards, as with most other electronics…

SD cards will inherently have less endurance than an SSD, so do keep backup’s of important data.