Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> who cares if it can store an exabyte if it takes all month to read it

To be fair, if I'm reading an exabyte in a month, my hardware's pushing >3 Tbps, which I'd be very happy with.



Plus just put 32 in stripping RAID if you really need to read an exabyte a day


Eh, that doesn't math out. It's the bandwidth per storage density (or ultimately per price) that matters.

If you have great cost per byte but your bandwidth per byte is bad enough that the price per byte doesn't make up for it then you have an issue.

They've started making hard drives with multiple heads because of this issue, they increased density to the point where it's not useful to continue adding density if it doesn't come with more bandwdith.


*RAED

Or maybe RAEND


RAVED is more likely. These things aren't cheap.


What is RAVED?


I read it as "Redundant Array of Very Expensive Disks".


But if you need 1eb, waiting a whole month for it isn't great. You'd be better off with 720 1pb devices taking an hour in parallel.


Yes it causes problems in this increasingly narrow situation.

Massive storage that takes a month to fully read is acceptable in a wide variety of use cases. If it's cheaper than hard drives it'll get a huge amount of users.


It's notable that 'time to read/write entire device' has been creeping up for any storage device you can buy off the shelf for the past ~40 years.

Reading a floppy disk took around 30 secs for example. A whole CD took 5 mins. My whole 1TB SSD takes 10 mins.


A modern hard drive (36TB @ 280MB/s) can take more than a day. If you treat a bank of tapes as one device this can get even more extreme.


Interesting, this is my first time consciously thinking about this trend.

Perhaps the needs for read/write speed are bounded (before processor, etc. becomes the limiting factor), while more capacity is only limited by price. Or maybe increasing density of storage inherently means a tradeoff with I/O speed (AFAIK, NAND flash needs to rewrite lots of data just to make a single write? Atom-scale interactions have side effects)


In long term archival use cases this is less of an issue. Especially if it’s many exabytes we’re talking about, needing to be stored for decades.

But I 100% agree with your main point about possibility vs productionisation.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: