Hello,
Was wondering if folks could weigh in on the ideal way to separate the different components of deadline?
1)Repo
2)Database
3)RCS
4)Pulse
5)AWS Portal/Asset Server
Should any of these be on the same machine? Which ones?
Should any of them absolutely be on their own machine?
And what type of machine spec would you recommend for each machine (VM or on-prem)?
You didn’t mention the size of the farm, e.g. ours is small at < 30 nodes, plus workstations added in when needed (we’re still on-prem). IMO, VMs or containers would be fine.
We have 1,2,4 running on one linux VM (vsphere) and 3,5 (RCS and AWS Portal Asset Server) on another, as we only use those when we spin up cloud spot fleets. On-prem render clients are windows, but we use linux for the ec2 instances / cloud nodes.
Repo is just a network share, so you don’t have to use a dedicated server for this. Just make the share volume minimal with space to grow and read-write access should be good (i.e. relatively fast) with the correct user permissions.
If you need replication or redundancy or HA or load balancing, then you can separate out and/or scale MongoDB and RCS.
Hi Jarak,
Our farm is very similar to yours: 18 on-prem dedicated farm nodes, about another 15 on-prem workstations that can join when not in use. Windows on-prem, linux ec2 instances via the deadline portal for cloud rendering.
All of our services reside on one machine at the moment except for the repo which is on a separate share with relatively fast speed. We will take your suggestions about separating the RCS and Portal into consideration.
Thanks,
-Jake