Has anyone deployed a small/mid-sized VDI using the Prox platform? And if so, what's your experience been?
I'm currently demoing a small cluster w/Ceph for static services at work, mainly license servers that require HA and it's going well so far. However, management is also interested in exploring a VDI setup to host remote and in-office thin clients.
With a limited setup, this seems very doable so long as the clients are restricted to low-impact usage (Win10 OS: office, email, web). My concern is scale in both number and capability if the deployment proves popular.
Compared to commercial alternatives, Prox does not have some key features native. One is a gateway/broker - but I have read some good things about Guacamole. Another is automatic live migration to load balance resources. And solutions for migration of PCI passthrough nodes (if possible, I know NVIDIA has a product for VGPU migration on VMware & Citrix).
Also are there any issues with using ZFS? Ceph certainly doesn't seem appropriate for this use case. HA wouldn't be required - periodic replication attached to a backup server would be more than enough.
Any advice would be much appreciated.
I'm currently demoing a small cluster w/Ceph for static services at work, mainly license servers that require HA and it's going well so far. However, management is also interested in exploring a VDI setup to host remote and in-office thin clients.
With a limited setup, this seems very doable so long as the clients are restricted to low-impact usage (Win10 OS: office, email, web). My concern is scale in both number and capability if the deployment proves popular.
Compared to commercial alternatives, Prox does not have some key features native. One is a gateway/broker - but I have read some good things about Guacamole. Another is automatic live migration to load balance resources. And solutions for migration of PCI passthrough nodes (if possible, I know NVIDIA has a product for VGPU migration on VMware & Citrix).
Also are there any issues with using ZFS? Ceph certainly doesn't seem appropriate for this use case. HA wouldn't be required - periodic replication attached to a backup server would be more than enough.
Any advice would be much appreciated.