The limits of SSDs

[Mercutio]

Found on Slashdot, reposting here.

I don't know if I care about long-term limitations of technology because engineers are great at improving stuff, but it seems that we're going to run into them headfirst.

 

[LunarMist]

Meh. Extrapolations like that indicate where not to go, not necessarily where it will go.

 

[Mercutio]

Well there are a couple worrisome trends. One is the focus on multi-cell memory rather than single cell and the second is the simple fact that smaller transistors mean reduced reliability. I understand that's the only way to make stuff cheaper, but both those things seem to run counter to the idea that this will be a stable storage medium.

Of course, we're also displacing traditional data storage with "cloud" systems now anyway. Maybe it won't matter by, say, 2018.

 

[Handruin]

Of course, we're also displacing traditional data storage with "cloud" systems now anyway. Maybe it won't matter by, say, 2018.

For this to happen, the ISP infrastructure in the US will need to be improved. There is a huge emphasis on cloud even in my own company, but I feel it'll be a bubble-burst unless the ISPs offer a cheaper, faster, and unthrottled service for getting to our cloud-based data in the next 7-10 years. It's getting there bit by bit, but we're so behind other countries with our affordable ISP service levels that its painful to get our data.

 

[ddrueding]

It would be interesting to have a "boot to cloud" device.

 

[Mercutio]

I'm more worried about bandwidth caps. For the most part the infrastructure to get decent speeds are available everyplace that cable is, and even DSL is technically a lot more capable than what's advertised.

Of course we're going to lag more densely populated countries for data service adoption; we don't all live right on top of each other like everybody in Europe and Asia.

What I can very easily foresee is a time when Netflix wants to send you a 10GB video stream and because Comcast hasn't ever in the history of time raised its data cap, you can only do that a few times a month.

 

[Mercutio]

It would be interesting to have a "boot to cloud" device.

Fast pipe plus something that will forward RARP to a listening server someplace on the public internet... it's not impossible to do now.

 

[LunarMist]

The cluod is for bulk data, not OS files and temp files. Latency is ridiculous.

 

[Mercutio]

You would load the OS image into local RAM. It's very possible to set up a modern *nix-based OS that way.

 

[Handruin]

I'm more worried about bandwidth caps. For the most part the infrastructure to get decent speeds are available everyplace that cable is, and even DSL is technically a lot more capable than what's advertised.

Of course we're going to lag more densely populated countries for data service adoption; we don't all live right on top of each other like everybody in Europe and Asia.

What I can very easily foresee is a time when Netflix wants to send you a 10GB video stream and because Comcast hasn't ever in the history of time raised its data cap, you can only do that a few times a month.

When I wrote throttling earlier, I meant the same as you with concerns to bandwidth caps. I agree that will be a big problem to overcome. Especially for people that only have one option for broadband internet. Decent speed is roughly available, but the cost isn't justifiable to the average consumer to make a cloud-based environment grow as fast as businesses want it to. You or I are willing to pay for the higher tiering of internet because we understand the benefits and have the appropriate budgets to justify it.

The example you explained is exactly what I'm talking about. ISPs (or in this case Comcast) will stint the growth of cloud-based services in the US.

 

[Handruin]

It would be interesting to have a "boot to cloud" device.

I know this isn't exactly what you're looking for, but Bitcasa is getting closer to this model. I have a beta invite for it but haven't tried it yet. I can see if I can more invites if you or anyone else is interested.

We could invent the boot to cloud device if one doesn't exist already.

 

[Will Rickards]

I always thought SSDs were just a transitional technology till they worked out RAM that saved state?