Slashdot Mirror

I'm thinking the middle end has dropped from $180 to $120, but just gut, no real evidence.

Well, here is an attempt at real evidence. http://www.littletechshoppe.com/ns1625/winchest.html

I think the next decade is a little ambitious. The problem is the amount of space required to store the data, which is probably hundreds or even thousands of petabytes.

Disk storage follows a Moore's law of 12 months (100% increase every year): http://www.littletechshoppe.com/ns1625/winchest.html

From this, it follows that you should have a 1.5 Petabyte consumer-level hard-disk ten years from now, and break the thousand petabyte barrier at consumer level in a bit less than 20 years.

-We have TB of HD space for what $700-$800? It's not quite there, yet.

It's not quite 2015 yet, either.

Back in 2000 a 128MB Trek thumbdrive was $399. $3.12 per megabyte.

OK hard drives are cheaper. here is a nice historical table of the cost per gigabyte. For reference the number of Mb per $0.01c seems to go up by a factor of around 10 every 5 years. Meaning that 1 TB of storage should be around $70-80 in 5 years time.

And even then it won't be 2015 yet

Are you high? Because if you are not, I must be in heaven paying only $80 for a 160GB drive, when according to your "theory" you must have paid out of your ass for a similar size drive.

In short, hard drive prices do go down, a lot. Especially considering that MS never upgraded the Xbox drive from the tiny 10GB hard drive it originally came with, it wasn't a significant cost.

Wow -- I just wasted a ton of time reading some of the links from the link you gave us. I'd especially recommend the History of Telegraph and Telephone Companies in Nova Scotia -- Thanks (and well done to the author)!

Someone else pointed out that the price of computers never really change, but that there is more power for the same price. In 1987 our family computer (mid-range) and printer cost around $1200. Today the same amount of money will also buy a mid-range computer (at least for gaming). However, this idea is getting less and less true as computers become commoditized and "powerful enough".

A few people have made the comment that Google can do this because 99% of the people will only use a few MBs of storage anyway. Reasonable theory, but here's another idea -- it doesn't matter if everyone uses a massive amount of storage.

First, figure out how many people there are in the world that might potentially use Gmail. Then figure out what is the potential maximum amount of unique data each of those people could generate on a daily basis. Then determine the size of the redundant information that could pass through the Gmail servers.

Note that a huge percentage of emails and attachments are sent to multiple recipients. For each piece of email or attachment compute and store a unique hash. Each account consists of only a list of hashes and some header metadata. This redundant information will significantly reduce the total storage space.

A quick seach finds this Berkeley study that suggests that there were about 400 PB of email (unique) generated last year. Assuming that you can save 1 GB of data for the fully-loaded cost of $1 (US), storing all of the internet's annual email traffic costs $500M annually in the worst case.

The best case is significantly better than that, as you can:

a) compress text by up to 80%
b) store every mail only once
c) store every large binary only once
d) add storage as needed, not up-front
e) reduce the cost of storage over time

This is off-the-cuff, but Google is looking at maybe a $50M annual investment in storage to store all the email on the internet, even if everyone uses it. They don't even need a storage limit. Period.

Not a curve, but I have two relevant bookmarks, both found while looking for something else: Historical Cost-of-Storage Data, and a great article about trends and such by a storage engineer at SGI. cheers,
mike