I'm likely to move a server from the data centre to my home office as I'm fortunate enough to have fast connection to the Internet. Generally what is the preferred option; A dedicated server A vm? The only preference I can think is that the vm would allow for easy snap shots should anything go wrong.
Hi Toucan, From my experience I believe running virtual nodes has great benefits over running just a hardware node, but it depends on your long term goals. Benefits of running a single hardware node: - Processes have access to all system resources - Only configuration of the hardware node is required Benefits of running one or more virtual nodes on one hardware node: - Can access all system resources allocated to it - Resources can easily be adjusted to accomodate new server expansion - Easy to backup - Easy to restore - Hardware node crashes, easy to move to another server within a short period of time I'm sure there are more, but that's a start. I use CentOS on my hardware node with OpenVZ. Once Falko pointed me to vzsplit, it was trivial to configure the virtual nodes. I run Debian Squeeze on the virtual nodes, and everything runs great! Before, I tried using squeeze with bean counters on the hardware node. What a pain... It was difficult to configure. Now, I just use swappages and physpages, which makes a lot more sense. I hope this helps generate more ideas on your decision. Darin
Agree with Darin's answer. I'd never use non-VM setups unless there is a total need for it (no idea what this could be).
Thank you both. I can definately see the advantages. There's definately times I wish I could have returned to a snapshot of the week before. I only have one external ip, so I assume I'd use that for the machine and an internal ip for the host machine? I think it's time I looked up some tutorials and start experimenting!
This week was the week I was grateful for you advice so many months ago. Although blessed with a blistering fast internet connection, my hardware has it's limitations. Having realised something was going horribly wrong with the server hardware, I stopped the VM, exported it to my laptop cloning the same mac addresses and had it up and running within a very short time. I took out the offending memory from the server, had it back up and running during the night and reinstated the original machine. AND NOBODY KNEW ANY DIFFERENT! yes... if I was being very cool could have used teleport to absolutely minimise down time; but this method was good enough. Let this be a lesson to all! I run a hardware host node of ubuntu headless with virtualbox, with two debian squeeze vms - one as the master ispconfig interface and billing module, and the other as the slave web server. thank you for the advice!
