The piece I'm currently working on will be 40-50 minutes long, written for a chamber ensemble of about 20 instruments. I am wondering what best practices/tips there might be for managing this project, Sonar-wise. To keep file sizes small and my brain organized, I'm creating separate files for each section, most averaging 1-3 minutes in length. Most of these sections will overlap by a bar or two, however. (For example, while the piano holds a long chord to end Section 4, the clarinet that opens Section 5 starts up.) One of the features of this piece is that there are no breaks, everything all flows together. But I'm pretty sure I don't want a single Sonar project that's 50 minutes long and 20 instruments.
 
1. I assume simply writing down slider levels and making use of FX chains is the best way to ensure the instruments sound consistent between tracks, but is there some slicker way? What if, in section 4, I've decided the clarinet is too loud, and I want to bring it down. Then I really should adjust all the clarinets in all the other projects, ugh. Anything I'm missing that could make things easier?
2. To do the overlaps between sections, I'm exporting each project as audio, then creating another project consisting of these audio exports, and overlapping these appropriately, then exporting that new project as a single audio file. Is there a better way?
 
Any tips? How would you go about this? I'm only 10 minutes into this piece so far, so I can change methods, but it's not like I haven't already done a lot of work on it.

jkoseattle
1. I assume simply writing down slider levels and making use of FX chains is the best way to ensure the instruments sound consistent between tracks, but is there some slicker way? What if, in section 4, I've decided the clarinet is too loud, and I want to bring it down. Then I really should adjust all the clarinets in all the other projects, ugh. Anything I'm missing that could make things easier?

You could use Clip Gain envelopes or volume automation.
https://www.cakewalk.com/Documentation?product=SONAR%20X3&language=3&help=Tools.22.html#1378431
 
Other than that, I would be making use of markers to quickly figure out where I am in the timeline.
https://www.cakewalk.com/Documentation?product=SONAR%20X3&language=3&help=Arranging.30.html
 

Are you recording live instruments? I've done some long recordings just fine. Just make sure you play safe with the levels, buffer size and reboot the computer first and close any processes that aren't necessary just to mitigate the chance of something technical going wrong. You want to avoid that embarrassing walk of shame from the control room to the live room, waving your arms about, shouting, "Hold it hold it!" 49 minutes in. Then having to explain that something dropped out, froze, crashed, peaked etc. I don't remember ever having a problem like this during a long take though. Although of course, I have joked about it after the performance. "Sorry, blew the levels 15 seconds into the intro. We'll have to go again." Or even, "Sorry was that supposed to be a take? I was just doing levels. I didn't record it." Just to see their faces :)

I would not split up front (you could always do that later if you need) but IMHO you will be better off having it all in one project. Use markers heavily to navigate. Get a good overall balance for the entire piece. Then fine tune using automation...

I'm with Rob. It seems unhelpful, to say the least, to split the piece up front. I've never had a problem with pieces that long or with that many instruments (e.g. live recordings of bands and then editing, overdubbing etc.). Just make good use of markers.

Sonar handles audio by spooling it to disk as recorded, then played back direct from the disk. So how long a recording is shouldn't make any difference, the restrictions on e.g. the number of tracks or plugins the PC can handle are exactly the same for a 4 minute song or an entire symphony.

Bu the sound of things you are scoring/arranging using samplers/vstis, not recording 20 instruments? Is that right?

If so the only reasons for chopping things into small chunks I can think of are firstly to make navigating around the work easier - 40-50 minutes is a long timeline to move around, and in case of a crash in Sonar or the PC. The latter is best guarded against by saving early, saving often, and using versioning, autosave and creating new complete projects every so often by using "save as" so there's a continual process of ensuring an (almost) up to the minute version of the project is available on disk.

And don't forget to back it all up to another disk every so often and when you stop - losing that much work to a disk failure or something would not be entertaining.

Thanks everyone. I knew I forgot something. These are all EastWest sampled instruments, for now. A few will be live players down the line (a favorite trick: insert 1 or 2 live players into a synth ensemble to make the whole thing sound live!). I guess another reason for keeping it chopped up was to be able to export versions and hear them away from the computer to take notes.
 
I'm surprised the vast majority of you are saying to keep it all as one huge file, but I admit that does sound appealing. I think I'm worried about navigation mostly. Markers are such a limited feature. I would want to use them both to mark whole sections, and I also use them when I'm scoring a bunch of instruments over a complex chord progression, I make a marker for each chord change. I wish they'd give us multiple levels of markers for that. 

Fudge for multiple markers (neat idea by the way): add one or more dummy (muted/archived) MIDI tracks, put a short note where you want to navigate to, and then use tab/shift-tab with that track in focus to jump to next/previous note. Obviously, the more tracks, the more flexibility...but the more chance you have of getting confused :-)

I like that idea of using dummy midi tracks.

But I don't really understand why you would need markers for every chord in a progression. Would it maybe be enough to mark the beginning and end and then use a dummy midi track with the actual chords and when they change?

Sanderxpander
I like that idea of using dummy midi tracks.

But I don't really understand why you would need markers for every chord in a progression. Would it maybe be enough to mark the beginning and end and then use a dummy midi track with the actual chords and when they change?

I might have a passage that has a dozen chord changes over a few bars, not unusual at all for my work, and those chords are messy things like Fsus7-9/Eb or whatever, and then I'll go into PRV for just the winds and arrange them across those dozen chords, then I'll do the same thing for the strings, and so on. Then when things don't sound right I can go back into PRV and look at what notes are being played and compare that to the actual chord in the marker that it's supposed to be. Such as "Oh, this is a Fsus7-9/Eb, but there's an A in the bassoon, that's wrong. Without those markers, I get super confused. I could do it with a dummy track simply playing the chords, but I still like to know what the chords are by name, in case I need to change them.
 
Hey, I like the fudge, I've never used notes, I'll check that out, and if tab/shift-tab moves between them, that's a rare case of a keystroke I'm not already dedicating to some other task! Thanks!