If it were on the servers, I suspect they'd have solved it by now. Those are under their direct control, are far easier for them to instrument/debug, and are fully characterized, fixed-specification environments. It makes far more sense that a persistent problem that's been so hard to track down is in the client, and it's great to hear they've added more instrumentation to those. Sounds like they need even more.
It's far worse on the server end due to the N-squared nature of the communications they have to do for players who are near each other. But like I said, that's well-characterized... they know what the servers can handle. LOTRO's architecture used to permit large, open worlds means that unless they're using layering tech in an area, they cannot spread the load across multiple servers - at least within a single 160x160 meter landblock. That's why this game will never have 100 players in a small area battling each other. N^2 is getting quite large at that point. Games like GW2 were architected differently. For example, if you're going to an event with 200 other players it can actually handle that. Sort of. What it does is to shrink the radius at which characters can get data about each other from something pretty typical for MMOs (like 100 feet) down to something truly tiny. I'm talking 20 feet here. Past that radius, the hordes of players just "vanish" if there are too many of them. Pretty weird, actually, but it does keep the servers from falling over.Anytime the number of people in an area increases, so does the consumption of bandwidth, on both ends.
Anyway, clients are far trickier to debug. The developer doesn't control their environment (which varies immensely), has fewer tools to work with out there, and often cannot easily reproduce the problems seen by customers in real time. When I've had to deal with stuff like that, I've done what they seem to be doing... carpet bombing the critical sections with performance logging, shared resource logging, etc, all in the hopes of getting enough data to figure out plausible causes for whatever multi-threaded nightmare of a problem I'm dealing with.
Khafar