Have you had experience implementing transactions spanning multiple nodes or systems?

Yes, experience in implementing distributed transactions that span multiple nodes or systems is usually associated with the use of two-phase commit (2PC) protocols or more modern approaches such as Saga.

In Java, this is often implemented using:

• JTA (Java Transaction API) with support for distributed transactions through transaction managers like Atomikos, Bitronix.
• Using XA resources to coordinate transactions across different databases or systems.

Example: when updating data in two different databases on different servers, the transaction manager coordinates the preparation and commit of changes on both sides to ensure atomicity.

It is also possible to implement this using microservice architecture and the Saga pattern, where a transaction is broken into a series of local transactions with compensating actions in case of errors.