Weekend performance test with Cashu nutshell with sqlite backend and 11 clients constantly swapping. didn't measure but 10 tx/s feels about right, but I haven't started optimizing the database access yet.
most of the cpu goes towards the 11 clients running on the same machine so this isn't a good test for neither the mint's performance, nor the client's :)
Is there something I lost about cashu design or there's a serious teorethical problem of DDOS?
Even if the cashu-swap is super optimized, how can a mint prevent ddos and still keeping the swap-fees at zero and without invading user privacy/UX with captcha et similar?
fees
so the magic cash-like experience of "offline tx without fees" will soon become "offline tx at 1sat-fees" 😣
yes and that's totally fine
there could be a way to implement a client-side PoW integrated with the cryptographic process of the swap? Maybe with a difficulty controlled arbitrarly by mint itself. In the end theres a lot of trust in mint to not rug and to keep their services online, adding a pow requirement to raise on necessity (or raise slowly with years as computing powers for users increase) dont change nothing in that sense.
Keep the pow requirement something like that a current gen smartphone can solve in 1-2 seconds could be enough.
Just a conceptual example, I dont understand nothing of this shit
Today(as I undestrand): mint swap your ecash if you send to it the valid piece of data.
Tomorrow: mint swap your ecash if you send to it:
{the valid piece of data} + {sha2 of [(the valid piece of data) + (the lastest bitcoin block ID) + (arbitrary data)]} + {the arbitrary data}
we did that but it turned out to be impractical. the difficulty is either too high for phones or too low for servers. pow is useful for Bitcoin but extremely impractical for many other applications.
