Slow and inconsistent suspend and resume times

[codeBuilder2011]

Hi

I'm trying out Vmware Fusion as a replacement to Parallels.  I am used to speedy suspend and resume times in Parallels - with Vmware Fusion both are generally slow and quite inconsistent in the time they take.  I have spent a while searching the Internet and this forum and just found years of different posts all suggesting the same issue.

My setup is a MacBook Pro i9 with 32GB RAM.  The VM is a Windows 10 machine with 8GB RAM and 6 cores.  I've tried dropping this down to 2 cores and it makes no difference.  Vmware Fusion is 11.0.2 and it's running on macOS 10.14.3.

The only thing I've found that I can't change is some people reporting this issue with APFS volumes (which I'm using) but if that is a problem I'd expect VMWare to make that clear in system requirements.

I'm hoping someone maybe able to help me as generally the actual running performance of VMWare is comparable and in some ways better than Parallels.

Thanks in advance for any advice!

Mike.

[wila]

Hi Mike,

When you say:

with Vmware Fusion both are generally slow and quite inconsistent in the time they take

Can you give us any numbers? Are we talking seconds, minutes, more?

Slow is a very relative number.

Re. the issues with APFS, those are supposedly solved. But perhaps there's an edge case that you are bumping into.

You can actually test that by creating a HFS volume on top of your APFS file system. (Disk Utility -> File -> New Image )

Make it large enough to contain your VM + RAM + additional size for snapshots.

Then copy your VM into that image. (yes, copy, not move as you're experimenting. Note though that you should answer the question "did you copy or move" with move not copy as otherwise you will have to reactivate Windows 10)

--

Wil

[codeBuilder2011]

Thanks for the reply.  Suspends take anywhere up to 3 minutes.  Resumes are much worse and can take well over 5 minutes. 

I've tried moving the machine into a sparse bundle and it made no difference unfortunately.

[ColoradoMarmot]

One note on Wila's image option - I used a sparsebundle format when I did that after APFS came out, and it made a huge difference with suspend.  Most of that difference is gone, but not all, and I strongly suspect this behavior is due to the migration of time machine to leverage APFS snapshots, which gets particularly bad if you're running disconnected from the time machine backup drive itself.  You could try disabling time machine completely (and if you use SuperDuper or Carbon Copy Cloner making sure that those snapshots are disabled as well),  rebooting the machine, and then testing it.  So far, mine have been inconclusive.

Aside, It won't make a difference for the restore, but with 6 you're likely starving the host.  No guest should have more than N-1 cores assigned, where N is the physical cores in the system (in this case, 6). 

[codeBuilder2011]

Time Machine is an interesting one as during the day I am away from my Time Machine backup drive.  I'll give that a whirl and see if it makes a difference.

And noted on cores - and to be honest, dropping to 2 hasn't made any real performance difference once the VM is up and running - it's still pretty good.  It's just this suspend/resume that's annoying.