Signing is a complex process, but perhaps reducing time is already on the roadmap?
For L2 Ethereum flows, 12+ seconds feels excessive compared to the UX in non‑IC apps.
Besides that, it costs 0.03$ to sign a transaction, which is very costly for high-volume apps. I suppose it is going to be lower if the delay is reduced?
Hi @dantol29, thanks a lot for your interest! You should be able to reduce the latency substantially by deploying the canister calling the sign_with_ecdsa endpoint on the signing subnet directly. In order to avoid cross-net latency entirely, the same principal applies for other canisters on the critical path of your application. For the production keys, the signing subnet is the fiduciary subnet (pzp6e).
In addition, DFINITY is currently working on increasing the throughput of all threshold signing keys significantly, which will also result in a proportional reduction of cost.
Thanks for the reply, glad to hear that it’s being worked on.
Is it possible to migrate an existing canister to a different subnet so that key signatures stay the same? A lot of users rely on current generated addresses
This will be made possible by the upcoming canister migration feature (https://forum.dfinity.org/t/canister-migration/41585). The work on this is in progress as part of the “Fission” milestone. Recently, proposal 138487 was adopted to initialize the new canister migration canister on the NNS subnet.