Thomas Eizinger

We can start with using this: https://docs.rs/mtu/latest/mtu/

So just after new install? I've read that notifications with Tauri are sometimes buggy because the application needs to be properly registered somewhere. Maybe Windows only does that on the...

We could write them locally into an sqlite DB and send that up to the portal every now and then? That would likely mitigate dataloss when the GW shuts down...

We should consider taking a similar approach as prometheus does: Exporting metrics just shares the current state of the counters but doesn't reset anything internally. That makes things more resilient...

The association with a _flow_ might be a bit tricky and confusing here due to the potentially overlapping nature of our policies. Can we just associate it with a client...

> After chatting with [@bmanifold](https://github.com/bmanifold) it would also be helpful if we could separate these from direct vs relayed bytes so that we can also tally this on the backend...

> > After chatting with [@bmanifold](https://github.com/bmanifold) it would also be helpful if we could separate these from direct vs relayed bytes so that we can also tally this on the...

> The association with a _flow_ might be a bit tricky and confusing here due to the potentially overlapping nature of our policies. Can we just associate it with a...

Another question to answer is: When do we do the accounting of return packets from resources? Do we count all packets? Only the ones that pass through the filters? Only...

> For the above, I'm sure we'll need to make sure we use lock-free data structures if possible and perform any aggregations possibly in a separate thread, right? All our...