7

I don't think it is possible to give a proper answer to this question, as we are lacking benchmarks from high-end hardware. I will, however, try to estimate a value: My GPU, a Nvidia GeForce 730 takes 24 seconds (2,5/second) on average to attach an address to the tangle. I measured this by using the light wallet on an idle Windows 10 PC. I started the ...


6

Spam is only good, if it is widely distributed over the network. So if there are enough fullnodes which are used for the spamming transactions, everything is fine. If only a few fullnodes are spammed, this could harm the network. Additionally there is the coordinator, which is a bottleneck right now. The devs are using spamming attacks, to research the ...


5

The main two bottlenecks on the tangle are as follows: 1) The coordinator: currently, the tangle has a coordinator that ensures that the network isn't overwhelmed by a malicious attacker. Due to the coordinator, there is a limit on transactions per second which can cause slowdowns. Once the network is mature enough, the coordinator will be removed and ...


3

The protocol does not enforce a particular tip selection algorithm, so every transaction-issuing network participant is free to choose whatever method they want. The IOTA reference implementation (IRI) has chosen MCMC. As demonstrated in Equilibria in the Tangle from https://arxiv.org/abs/1712.05385 all network participants should (in theory) converge to the ...


2

The highlighted points of the tangle whitepaper are that first, you cannot enforce a selection strategy, and second, if you assume others are following monte carlo, then that is your best strategy to follow in order to be selected by others.


1

The IRI implements a random tip selection algorithm but you are free to modify it to select again and again the same transaction. It's what you do when you promote a transaction. Inundating the network should be very costly because you have to do the pow for every transaction.


1

Trying to recall something mentioned by CfB a while back, I believe IOTA will eventually use collective memory usage: In a sense that hash-based signatures will be reduced from 8192 to 64 bytes - which is over 99% compression therefore size will not be an issue. Also don't forget about snapshots


Only top voted, non community-wiki answers of a minimum length are eligible