I think this discussion will work better if we can leave the judgement of personal background and qualification out of it.
This discussion was degraded by dozen of completely useless, judgmental, and clueless demands based on unrealistic expectations.
It is NOT irrelevant to know who are you dealing with.
It places statements in the context.
Person in question never came here to ask for the advice what to buy, with set of expectations.
As I said, if he/she/it/they did, I would recommend:
1. Vector or Kvaser CAN analyser. This is tool that exist for this kind of work
2. Some kind of streaming USB analyser (Saelae or similar). Also specific instrument for capturing long sequences.
3. Deep memory Picoscope. They are very popular in car industry for this kind of work, on CAN specifically. I did a test on my 3406D with 500kbit CAN and could capture 29000+ packets in 10 sec period with no problem. There is export to CSV and Excel. Make note Picoscope is specialty instrument, 3406D has 500MPts, processing is done on your PC, and my I7 CPU and 32 GB RAM PC took 20+ seconds to decode that single capture. It cost 2500€ for Pico alone, no PC price included. But it worked.
No embedded scope I know that is not some 20000 USD can come even close to what eTobey expects on one of the cheapest scopes from Siglent.
My Keysight barely captured 1300 something packets before bombing out. That was 12000 USD MSRP scope at time of purchase few years ago.
So I have zero tolerance about "its easy, why is everybody so stupid, I demand....."
And also you:
How do you know 2000 CAN decode messages are orders of magnitude below memory limit?
Did you notice some other protocols can have more, in order of tens of thousands, but CAN cannot?
Maybe decoding CAN is more memory/resource intensive than what you think ?
Hint: eTobey speaks about memory needed to store results. How much memory is needed for process itself? Would it be possible but with specific optimizations that would go against how memory management in scope works for other things... Why other scopes also have limits on this that are much less than 20000 that is "easy...
Why is it "expected" that 500€ scope that already does better than 12000 € scope (2000 as opposed to cca 1300 on MSOX3000T) should do not only 50% more that it does now, but should do orders of magnitude because otherwise it is stupid??
What kind of thinking is that?