Using VEP as I do ( a single 64b server / same machine) does result in better performance overall.
Can't tell you in tech terms but it does. I wouldn't try to do any orchestral projects w/o it now. And I have done them in both Sonar & Cubase previously.
I launch VEP once in the morning  (with all my orchestral sample libraries loaded) & work on different projects in different daws, open & closing them all day long until I turn off my puter at night.Works great for me, I hated loading all my sample libraries over every time I opened or closed a project.

BTW,
 
I just added a tutorial video on using VE Pro with Sonar. If you are interested in using VE Pro with Sonar it might be helpful.
 
http://www.daveranck.com/#!tutorials/c1ajd
 

Thanks - thats superb!! Will look at it later tonight

Thank you . Excellent tutorial.

wish I'd seen your video tutorial before today! Very helpful! Thanks!!

DRanck
I recently completed a composition in Sonar and for the first time, hosted the East West Play instances in VE Pro 5. I 
 Just thought I'd share my experience. Maybe it will be useful to someone.
 

 
Thank you for sharing this; I'm thinking about making the leap to VE Pro 5 myself.   I'm very glad to hear it works well in Sonar since I really didn't want to move to Cubase. 
 

DRanck
UPDATE:
I made 2 changes that make working with VE Pro even better:

1. I am now working decoupled. This means that the VE Pro project does not load automatically with Sonar. This is working so well (in conjunction with #2 below) that I saved the VE Pro project as a Metaframe in the same folder as the cakewalk project. This will now be my starting point for new orchestral projects. Since the Metaframe is decoupled I can load it once in VE Pro and then switch to another Sonar project that uses the same template wihtout reloading the samples. Great time saver. 
2. I moved the Hollywood Orchestra samples to a USB 3 SSD. Load times are WAY shorter now. Big difference over 7200 RPM internal mechanical drives.

Depending on your polyphony needs, you have numerous good options for sample drives.

• If you need lots of polyphony and large storage space (and your virtual instrument can only address a single drive/location), you can put a pair of SATA-III SSDs in RAID.  That'll net 1000MB/Sec.
• If you need massive polyphony, you can install an M2 Ultra SSD that uses 4 PCIe lanes (2500MB/Sec).  The only downside is that you're (currently) limited to ~500GB size.
• If you need massive polyphony, you can install a PCIe x4 SSD (uses 4 PCIe lanes).  Sustains 2500MB/Sec.       A 1.2TB PCIe x4 SSD goes for ~$800-$1000.

If you're an extreme example, you can use a combination of SSDs to achieve disk-streaming polyphony of 4000+ stereo voices.     That should cover most anything...   
 
FWIW, if you pulled that USB3 SSD out of its enclosure and installed it internally via SATA-III, it would perform even better.
External USB-III SSDs usually sustain ~360MB/Sec.  If it's a good SATA-III SSD, it'll sustain ~520MB/Sec.

I don't understand why running VEP on the SAME machine would result in smoother operation, lower latency settings, and higher track counts... wouldn't it all be the same, if not greater (VEP itself) load on the machine? Does VEP handle instrument loading/cpu management differently/more efficiently than Sonar?

 
Simply put, VEP and Sonar run in separate processes and have their own thread pool. Sonar has a lot of work to do besides managing the VSTs. VEP really has one job - manage the VSTs. By running in separate processes better usage of system resources can be achieved. Divide and conquer.

DRanck
Allan - I have one instance of Play for each section of the orchestra. Each of those contains 1 - 7 articulations. Each articulation is driven from a single midi track in Sonar. I find it easy keep things organized that way. And EW recommends multiple Play instances. Even if I wanted to, I can't fit everything in one instance of Play. Also after reading EW forums, this seems to a common configuration.

Dave - I am fairly new to this but I've been working the past week or so on all things related to Sonar/VEP5 Pro /Articulation Tracks vs Key Switches, etc.   I have many libraries and I finally decided to go the VEP route, but in setting it up, as I answer questions, I find more questions about which best practice to use (I understand that different methods work best for everyone).   On paper I could see ending up with 1000 tracks, if I use a separate track for each articulation for multiple string and orchestral libraries.  I don't want to build and rebuild a template 100 times as I learn new things....if I can help it.   So....
 
I have many questions but one stands out right now..... what is your method for using ONE Sonar Midi Track to trigger multiple articulations in PLAY? 

I have many questions but one stands out right now..... what is your method for using ONE Sonar Midi Track to trigger multiple articulations in PLAY?

 
 
The only time I use multiple articulations on a single track is when I'm using a keyswitch patch. I can't imagine why you would need 1000 tracks! There are maybe 18 - 20 sections in a normal orchestra. If you look at the powerful system template for strings there are about 11 articulations that cover almost every eventuality. Add a few more to cover some special circumstances and that would be 15 articulations. And that is a lot. I usually don't actually use nearly that many per section. Even the pros on the EW forums seem to use around 100 tracks.
 
Why do you need so many tracks?
 
Dave