Fork VMs & Plot Access Times - Network share or Local Machine Harvesters?

Hi guys. I’m running chia on my local machine and have my plots in JBod storage racks.

I have a load of forks in their own VMs through windows hyper v.

It’s a dell server I’m using for farming and the fork VMs use virtusl switch connected to a different ethernet port than my local chia machine.

Currently the forks access my plots through network file sharing and due to being on different ethernet ports, they must be accessing over my lan. Plot response times can vary from 2s to 9s sometimes. I’m guessing this set-up is creating a lot of traffic on my lan which could be causing bottlenecks.

I’m going to be adding a lot more forks soon.

Would I be better setting up individual harvesters on my main local machine, which has direct access to the jbods, instead of the all the different full nodes accessing the plots via network??

Would this reduce network traffic and perhaps speed up my plot access times?

Due to the rediculous number of vms - I’m in the process of changing my farming machine to quad 8 core xeons, so 32 cores, 64 threads and RAM isn’t an issue so running all these vms and harvesters shouldn’t be a problem. At the moment, vms are running of raid 0 arrays on 15k sas drives - will upgrade to ssds later but I don’t think it’s causing any issues with speeds there.

Or is there another way I don’t know about?

Just want to make sure I’m not gonna be missing out on rewards because of a bad configuration. Networking, vms etc is all pretty new to me so learning as I go.

Thanks in advance.

Sorry I forgot to mention - the more Fork VMs I have running, the slower my access times become.

Harvesters are always faster than network shares.
So I’d assume you will fare better without the network shares.

I would certainly imagine so

1 Like