Roon uses the same database resources whether the file in your library is local or streamed. Once you’ve added something to your library, the database creates/tracks all the associated objects.
I am not sure if there is any “overhead” because you have different versions in your library. I rather doubt it has much, if any, effect.
Keep in mind Tidal has a hard limit (10k) on number of albums you can favorite.
Nice one Mike - i will take a look at that.
Oooo! You know I love Foobar!
Maybe I’ll give it another try, but a few months ago it crashed quite frequently.
Thank you.
Very interesting read Michael.
I might come back later, but I glad I stirred this pot again and gained comments from you and others.
One of the major problems for many applications is the length of the file names. You have to consider that application scans not only file name but also the path where the file resides. Some applications may throw an error message, but others may supress errors and continue scanning. There might be also issue of file permissions that prevent application from reading files. Roon stores all relevant data in its database. Depending on its size it could take longer to filter out request. However, from what I see Roon is optimized for data searching and retrieval. Under no circumstances application should store data temporary in memory etc. It is all about queries. in this case CPU and RAM matters.
I suspect long term collectors fall in the 100-200 k tracks range. I guess there’s quite a few of those out there. There was a thread a while back “how big is your library”
Some stream only with minimal libraries, one user listed 20 k albums and a million tracks the other day, but these are rare
For example mine is reported by Roon as 130 k tracks
For that my core is a tower PC , i7, 7700, 16 gb RAM ,250. SSD it handles it no problem
PS Tidal I believe has a 10000 limit 
I’ve seen this statement before, but surely there must be differences? I mean Volume levelling data and Audio Analysis information is not stored for online content as examples?
Or is that a separate part of the database perhaps…
cuetools will be equally good for converting single flles with cuesheet to separate files. So you don’t have to use foobar2000. 
I only mentioned, because some folks around here already use it for other things. (p.s., I’ve been using foobar2000 for over 10 years and it’s one of the most stable pieces of software I have…I’ve never had it crash.)
edit: Mostly I’m just embarrassed I didn’t think earlier about the issue of “single files with cuesheet” as a reason that you weren’t getting the same count of files in Roon, as Roon doesn’t recognize those files.
Good points! Thank you.
I have large library. I noticed my server always froze up when scanning a particular artist. The files didn’t show in the skipped files section. As soon as I deleted that particular folder, the scan of my large library worked fine, it is over 200,000. I couldn’t figure out why it wouldn’t scan that folder. The files played fine using a different player. I plan to move one album in at a time, to see which particular sub folder is the issue.
I presume that my library of 665,196 tracks (including 8561 Tidal tracks) would be considered large.
I use a Nucleus+ for storage consisting of a 4TB SSD and a 10TB external drive.
Using Roon on my 2 Macs, phone and Ipad could be extremely frustrating, with many crashes and laggy performance.
Since I upgraded the 8G of memory in the Nucleus to 32G performance has improved noticeably, but not as much as I would have liked. Search can be very slow and there can be a long gap when choosing another artist or album to play.
Overall I’m satisfied but I’d apreciate any tips that would improve Roon’s performance especially using search.
This may sound crazy, but one way would be to set up 2 totally separate Roon systems. For instance, put all your classical on one and all your non-classical on the other. Purchase two Roon licenses. If you can afford almost $1 million worth of music, two Roon licenses should be a non issue.
There’s a point to made here about the quality of a network affecting user experience of Roon.
My library would objectively be deemed huge yet from a performance perspective my user experience is generally good running the core on an i5-7600T CPU @ 2.80GHz with 16GB of RAM that’s pretty slow by current standards.
My WiFi network is, however, 100% Ubiquiti based and there’s no mesh involved, with each access point wired into a gigabit switch and only internet bound/originating data passing through the ISP router.
It is actually a great idea. I can see how that would work.
However, Roon is already the most expensive bit of software that I use. I am not sure about the connection between the size of the liibrary and the willingness to pay more.
Upgrading the RAM was, I am sure, the right thing to do.
The processing power sounds pretty good to me! Sure, it could be improved but, for goodness sake, it is only a database.
The search is the worst thing. As you say, slow. But it was confirmed the other day when I searched for ‘John Hassell’ and it returned zero, that it is rather lacking. Surely fuzzy logic (or whatever is the right term) should allow searches to find ‘Jon Hassell’ or even accommodate ‘John Hasel’. But, no, a blank result.
I don’t understand what you are saying about the network … but it sure sounds impressive!
I saying nothing about search in this thread.
The i5 CPU in use dates back to Q1/2017 and it’s the T variety, hardly a performance CPU by standards back then and definitely not by current standards. RAM was 16GB from the outset, this is not a NUC or NUC equivalent.
Re network performance it probably explains (at least in part) the performance issues you speak of
Sorry, I seem to have invented that. Maybe I am replying to the wrong person.
But! Did I mention network problems? I don’t think I did, but it is such a long time since I made this post. I do not like wifi, I like wires!
The search is the worst thing. As you say, slow. But it was confirmed the other day when I searched for ‘John Hassell’ and it returned zero, that it is rather lacking. Surely fuzzy logic (or whatever is the right term) should allow searches to find ‘Jon Hassell’ or even accommodate ‘John Hasel’. But, no, a blank result.
you didn’t, and whilst I also prefer wires, when it comes to controllers (phones & tablets) they rely on wifi and in turn many users have wifi that’s reliant on their isp router. That’s where performance issues can quickly manifest.