Post
Topic
Board Speculation (Altcoins)
Re: [XMR] Monero Speculation
by
luigi1111
on 04/02/2016, 00:16:45 UTC
Even if Monero gradually and without penalty accommodates larger blocks, they will never be able to shovel 5k-120k tps onto Layer One.[/b]

That's a pretty big range. 5K is somewhat more plausible than 120k, for longer term.

5K can fit in a 100 megabit connection and require something like a large hard drive a month to store the blockchain. Not going to be home-based nodes at that point but it could still fit easily within satoshi's vision of many independent nodes in data centers, even small data centers (10 years is only 100 hard drives -- not really much a storage array). Ignoring future improvements in storage technology which I think is ArticMine's point.

Considering a peak 60x higher means you would need 6 gigabit connections, also plausible for a small-ish data center. As long as the peak isn't sustained too long, storage wouldn't become a problem, and verification can be almost arbitrarily parallelized.

I agree with your main point about there being better ways to scale than jamming everything into one chain though.

Small point: The block size can't grow at all without some penalty. Despite what the white paper says the penalty kicks in at the median size, not 10% above the median. The implementation has always worked that way.

I'll take that as confirmation Monero can verify 5K tps on something like a reasonably priced server.  What about 120k?  ~25 Xeon cluster?

Once crypto has defeated fiat and we all live in Libertopia, I'll have much less of a problem with full nodes residing mostly in data centers.

Of course there is every chance Libertopian homes will have full duplex 6 gigabit connections.

Thanks for nifty insight into the spec vs reality of block growth penalties.  It's nice to know Monero is protected from block bloating attacks, even if well-intentioned!   Cool

I did some thumb-sucking math a number of pages ago: I "built" a single system server that exists today (8-way) that could process 40-something-k TPS IIRC.

Edit: found post: https://bitcointalk.org/index.php?topic=753252.msg12768096#msg12768096

This would suggest to me that a 120k TPS node would cost ~$200,000 for just the system(s).

Assuming a transaction size of 2 KB (made up), we'd need a ~2gbps link.