How do you verify that channels behave the same after migrating off Mirth 4.5.2?

[jgere9]

I'm spending a few weeks researching how teams actually verify HL7 v2 interface migrations - Mirth Connect 4.5.2 to OIE (and to BridgeLink or the commercial 4.7) - and I'd rather ask people who have done it than guess from release notes.

The question is narrow: after you move channels to the new engine, how do you convince yourself (and whoever signs off) that they behave the same?

Specifically, I'd love to hear:

1. Do you run old and new side by side for a period, or cut over and watch?
2. If you compare output, what exactly do you compare - the encoded message, specific segments/fields, ACKs? And what do you do about fields that legitimately differ (timestamps, control IDs)?
3. What is not covered by comparing messages - routing decisions, error and retry paths, database writes, external calls - and how do you handle those?
4. Roughly how long does the verification step take relative to the migration itself?
5. Has anything ever slipped through and shown up in production? What was it?

Why I'm asking: "the channel imported fine" and "the channel behaves the same" are different claims, and between Rhino/E4X details, code template libraries and connector-level differences there seem to be plenty of places for the second one to quietly fail. I can't find a shared practice for demonstrating it, which either means everyone has their own, or that it mostly isn't done.

I'm not selling anything and I'm not looking for work. If there's interest I'll post a summary back into this thread once I've collected the answers here and from a handful of conversations elsewhere.