Scalability Costs
Scaling? Yes, towards thousands, ehm, hundred thou.., ehmm, Billions! We have tested our current commit-chain implementation with up to 100'000 client accounts, and the costs grow such that one commit-chain can support more than 1 Billion users 😊 - no matter how many transactions are done on the commit-chain.
Moreover, clients are light by nature and have only a few kb of storage and bandwidth requirements. Note, however, that clients are required to come online once in a round (every 36h) to audit their state. We plan to relax this requirements with watchtowers.
Now, please enjoy the following graph that shows on the x-axis up to 1 Billion users, and on the y-axis the blockchain costs for withdrawal (exit) and disputes with the operator.
More detailed and finer granular cost analysis can be found in the following table.
​ | NOCUST Costs | Cost Growth |
Checkpoint Notarization | 80k gas | Constant |
Deposit | 135k gas | Constant |
Withdraw Initiation | 150k gas, <400k gas @ 1 Billion users | O(log(n_users) + log(n_supported_tokens))) |
Dispute Initiation | 220k gas, <450k gas @ 1Billion users | O(log(n_users) + log(n_supported_tokens))) |
Dispute Answer | 250k gas, ~<200k gas @ 1Billion users | O(log(n_users) + log(n_supported_tokens))) |
Operator Storage | All client accounts | O(n_users) |
Operator Bandwidth | All active client accounts | O(n_users) |
Client Storage | 312 bytes per transfer in the last 2 rounds + 1kb | Constant |
Client Verification | Transfer signatures + one merkle proof per round | Constant |
Client Bandwidth | Own transactions only | Constant |
Fair Mass Exit | Yes | ​ |
Delayed Transaction Finality | Yes, within 72 hours | Free |
Instant Transaction Finality | Yes, insurance capital | Depending on client need |
Token Support | ETH, ERC20 | ​ |