Tracing an ILP Tx start to finish

I’d like to trace an ILP fiat transaction through the financial system, in order to determine its impact on the balance sheets of all affected parties.

The problem is I’m not sure exactly what is taking place when the asset being transferred is a bank deposit, as that asset is the liability of a bank, and that bank must use its reserve deposits on the Fed’s ledger to settle the transfer of its bank deposits to another bank.

From reading the white paper, it sounds like ILP creates instances or hypothetical ledgers for the parties engaging in the transaction, much like a payment channel on a distributed ledger, for example.

Is that the case? Are customers’ own ledgers (outside of ILP), the banks’ ledgers, and the Fed ledger, all unaffected by a transaction through ILP, and the “money” being exchanged is just a tally or point keeping system, and participants must settle their transaction through their banks separately, outside of ILP?

Today, yes. The way we generally think of ILP packets is that they create obligations that each pair of directly connected peers will track and settle on their own. If the peers’ chosen settlement mechanism is fast and inexpensive, like a payment channel, they may settle every packet or on a very fast schedule like every second. If the peers are comfortable operating with higher bilateral credit limits, they can batch up more packets and settle for higher amounts. (For more details, see this thread discussing what the “main” way of settling in Interledger should be)

That said, if banks and/or central banks adopted Interledger, they could theoretically integrate the protocol in such a way that the “point keeping system” is their actual ledger of record. However, given Interledger’s focus on extremely low latencies and high volumes of packets, it might make sense to use Interledger as a “clearing” system like a more open VisaNet and settle using slower systems that prioritize certainty over speed. People that have more expertise in traditional payment and settlement systems might be better able to weigh in on this topic than me.