• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Virtualize Render Farm?

U

/usr/sbin

Successfully Trolled by Megalith
Joined
Jul 18, 2010
Messages
3,927
My wife does work on the side that requires rendering. As a kind of geek hobby, I have three ESXi servers running at home (Sandy Bridge Era and later). These machines largely stay relatively idle. She wants to speed up rendering. I was thinking that instead of upgrading her PC, I could install a VM on each host for rendering, and modify the resource allocation so the render VM is something like 1 share, and regular VMs 50 shares each for CPU resources. Idea being that the render VM would only really use time when other workloads were idle. Everything is hardwired and attached to 48 port gig switch with 2 gig ports on each ESXi host hooked to the network as LAG.

Since I'm not an expert on virtualization, I wanted to check with some of you more seasoned folks and ask if there is a reason I'd not want to do this. The home VMs largely run low power tasks (DNS servers, privoxy, VPN server, etc.).
 
This is not a bad idea, however your approach to aligning the resources won't work as expected.

There are two parts to this--optimizing the rendering VM to run as efficiently as possible (assuming it's a multi-threaded application). In this scenario, you need to find out the best vCPU/memory sizing to achieve best performance. I would play around with different configurations to see what works.

The next step is ensuring that your "regular" VMs are not negatively impacted by the render farm. For that, I would suggest setting a CPU reservation for them (not shares).

Keep in mind that rendering applications can be very very resource intensive and will like impact your other VM's no matter what the configuration; fine-tuning the reservations and limits of the different resources can define to what extent.

Another approach would be to configure DRS affinity rules to pin the "regular" VMs to one host and the render farm to the other two.
 
I've got a 6 node render farm at work that runs really well almost as well as bare metal
 
You must log in or register to reply here.
Back
Top