I currently run a cluster of 5 gaming servers (and 3 sandboxes that are usually not running unless I need to spin up a backup) on 3 Linux boxes, and am looking to move everything to one vm host. I am a bit green when it comes to how docker and LXCs work and looking for suggestions (and hopefully explanations) of the best way to do this.
Each box has STEAMCMD installed, then I do three installations (in different folders) of my gaming servers. Each install of the game takes up about 20GB including save files. I then blow away the ~/ShooterGame/Content/ directory in install #2 and install #3. and symlink that folder to install #1. This drops the size of additional installs to 3-4GB each.
My goal is to virtualize and move everything to the same host, but in the most ram/cpu/storage efficient way possible. I like the concept of being able to individually allocate CPUs and RAM but those aren't deal killers, I have more than enough for what I'm doing.
I would love suggestions on the best way to move forward.
Do I:
1) Just P2V the boxes I have and keep things set up the exact same way?
2) Have individual VMs for all 5 (erm actually 8) gaming servers (this sounds like a huge waste of resources)
3) Use Docker? (It sounds like that would be just as inefficient as 1 or 2 with my current understanding of Docker)
4) Use LXCs - So now I only have one kernel instead of 3 in option 1 or 5 (or 8 when the sandboxes are running) in option 2... is that such a big deal? and is there a way to share the content folder between LXCs to save on storage?
5) Any other cool ideas?
Looking forward to suggestions and explanations,
John
Each box has STEAMCMD installed, then I do three installations (in different folders) of my gaming servers. Each install of the game takes up about 20GB including save files. I then blow away the ~/ShooterGame/Content/ directory in install #2 and install #3. and symlink that folder to install #1. This drops the size of additional installs to 3-4GB each.
My goal is to virtualize and move everything to the same host, but in the most ram/cpu/storage efficient way possible. I like the concept of being able to individually allocate CPUs and RAM but those aren't deal killers, I have more than enough for what I'm doing.
I would love suggestions on the best way to move forward.
Do I:
1) Just P2V the boxes I have and keep things set up the exact same way?
2) Have individual VMs for all 5 (erm actually 8) gaming servers (this sounds like a huge waste of resources)
3) Use Docker? (It sounds like that would be just as inefficient as 1 or 2 with my current understanding of Docker)
4) Use LXCs - So now I only have one kernel instead of 3 in option 1 or 5 (or 8 when the sandboxes are running) in option 2... is that such a big deal? and is there a way to share the content folder between LXCs to save on storage?
5) Any other cool ideas?
Looking forward to suggestions and explanations,
John