blockchain storage requirements

[zatoichi]

Anybody ever try to calculate the blockchain size when the last coin is mined? And the cost of that storage (assuming a network of a "reliable" size)? What kind of fee would be required to process one transaction at that time? Also if all current transactions globally were recorded in a blockchain, what would be the daily incremental increase in size?

[ScripterRon]

The block size is dependent on the number and size of the transactions included in the block.  So it is impossible to guess how big blocks will be in the future and thus how big the block chain will be.  There is currently a 1MB maximum block size, but who knows if that will remain the maximum as the number of transactions increases.  You could just take the current block size (around 250KB) and use that to guesstimate the future block chain size.

[zatoichi]

So nobody's done a "cost of ownership" study for this product?

[infinitybo]

@zatoichi Blockchain is interesting because technically it doesn't store cryptocurrencies but stores transaction information only.

[zatoichi]

I don't understand this. I thought the blockchain had "proof of work" information.

[Gabi]

https://en.bitcoin.it/wiki/Block_chain

[pontiacg5]

There is supposed to be a block every ten minutes, but they sometimes come along faster when the network is steadily growing. Right now, as already said, a block is generally nowhere near the maximum 1MB limit, but last I checked we were still around 6-8min/block.

So, today, you are looking at adding 36MB of data to the blockchain per day. At the most, you are looking at 144MB/day. There's already over 14GB (please confirm!) of data in the chain, so at some point "archiving" older years of the chain might be necessary.

But bandwidth is still relatively low, and storage space is cheap. Running a node is hardly a cost in the grand scheme of things, and it's not even necessary for normal users.

[zatoichi]

Do you know how many nodes there are?

[DeathAndTaxes]

I don't understand this. I thought the blockchain had "proof of work" information.

It does however the proof of work is in the blockheader.  The block header is only 80 bytes every ~10 minutes or 4.2 MB per year.
https://en.bitcoin.it/wiki/Block_hashing_algorithm

Of course that assumes an average 10 minutes per block, if the network hashrate is growing it will be higher so you could add say 30% to assume average 30% network hashrate growth over time and it still is ~5MB per year or 1GB every two centuries.

The overwhelming majority of the blockchain size is from transactions not blockheaders.  That size will depend on tx volume which is hard to quantify.  Currently their is a 1MB limit per block and the post above shows the max size per year under that limit but eventually that limit will either be removed or raised.

Still even that doesn't tell the "whole story" because the full historical block chain is not needed for either mining or verifying transactions, only the pruned blockchain which contains a copy of all unspent outputs (UXTO) is needed for that.   Currently the UXTO is about 10% of the full blockchain and as a % that will decline overtime.  Estimating the size of the UXTO requiring guesstimating the likely number of future users and the average unspent outputs per user.  The second factor is going to vary wildly depending on what usage scenario of Bitcoin in the future.  If Bitcoin is used primarily as a store of wealth (think digital gold) then the UXTO can be relatively small compared to the user base, and if it is a common high velocity transaction medium then the UXTO may be very large compared to the base.

So making any guestimates for something more than a century away is pretty much a shot in the dark.  The good news is that Moore's law is alive and on any century long timeline Bitcoin simply can not grow faster than Moore's law.  That means that over the course of a lifetime the relative "cost" will go down.  As an example a 6TBs of storage today costs less than a 1GB hard drive I purchased in 1995.  That is a 6,000 increase in storage per dollar in less than 20 years.   I have no doubt that in 20 more years a multi PB drive will cost less than a TB one will today.

[zatoichi]

Interesting to find out Moore's Law is applied to storage (I thought only speed). Speed of course has no brick wall, you can halve your distance eternally, but (ignoring quantum computing) if you pack a "transistor" into the space of an atom, you're pretty much done. Actually I take that back. You can only halve the distance between transistors until they are an atom apart. Same thing I guess as storage problem.

But I thought the mining algorithm only kept getting harder due to increasing blockchain size. So if you remove stuff it screws with the system.

[Gabi]

But I thought the mining algorithm only kept getting harder due to increasing blockchain size. So if you remove stuff it screws with the system.

False.

[zatoichi]

That would be a great true/false question for a bitcoin quiz. Wanna start collecting questions? I think a quiz score would be a good addition to profile information (rather than just "Activity" which is easy to get with one word answers....maybe they could multiply activity by post length).

[AldenHurley]

Interesting to find out Moore's Law is applied to storage (I thought only speed). Speed of course has no brick wall, you can halve your distance eternally, but (ignoring quantum computing) if you pack a "transistor" into the space of an atom, you're pretty much done. Actually I take that back. You can only halve the distance between transistors until they are an atom apart. Same thing I guess as storage problem.

But I thought the mining algorithm only kept getting harder due to increasing blockchain size. So if you remove stuff it screws with the system.

Not correct. Each block's hash acts as an input into the next potential block, creating a 'chain'. 'Removing' a block would invalidate all blocks recorded after it. That said, technically it's only necessary to maintain the merkle.

[Gabi]

What matter is that new people reading this thread understand what is right and what is wrong.

[zatoichi]

I think it is more important that they understand why it is true or false (knowledge integration). I still don't know why.

[Gabi]

I would have to copy-paste the bitcoin wiki to explain that, that's why i linked it. I know, it is not easy, but that's how it works

[zatoichi]

What's a merkle? would be another great quiz question (multiple choice). I looked on wikipedia and didn't get too far. Then I went to "hash chain". It seems that if you have h999(x) you can't get to h1000(x). Maybe. But if you have h1(x) you should be able to recreate. So you just need to get in at the beginning or at password change.

[zatoichi]

Went to the link for "block hashing algorithm". Pretty dense. I was interested in "difficulty" though and linked there. Apparently the difficulty can go down as well. That would be another great reason for introducing Bitcoin2 now. Not only would it be less difficult to mine Bitcoin2 now but the shift in computing power away from Bitcoin1 would reduce the Bitcoin1 difficulty. Maybe you would be able to go back and forth with greater profitability. Say, every two weeks?

[pontiacg5]

What?

Difficulty has absolutely nothing to do with block size, at all. Unless I am mistaken? It's simply a changing target, it always takes the same amount of integers to get there.

There already is a bitcoin 2, and a three, probably more. They are all complete failures as they should be.

[zatoichi]

You're right. I researched difficulty and found it can go down too. Depends on how long it takes for new bitcoins to appear. Bitcoin1 looked like a failure for a while (and may fail again...China knocked the wind out of it...and a vulnerability may be found eventually...worse than Sires/Cornell). Bitcoin2, being just a theoretical clone of Bitcoin1 would still have all the advantages of Bitcoin1 along with easier mining in the short term.