I know this is a pretty common theme these days with Broadcom just, well being who they are and wanting to plunder their user base. I support some very small companies, mostly non profits and VMware had a very generous pricing which was basically education that was really cheap for essentials, not essentials plus just essentials. But that is no more and the standard pricing from Broadcom well its a lot more expensive and not perpetual so looking at alternatives. Funny but all the alternatives including proxmox paid subscription is way more than the old VMware pricing.
Some background, will try to keep this short, I'm a linux guy for some time, so I understand what is being offered with prox but also looking at xcp-ng and some other options. I originally went with VMware because of cost and the if I get hit by a bus it would be easy for someone to come into and support. Currently we have 2 hosts running various OS, windows server and linux for the most part. Since we have essential we do not have vmotion but we have veeam that we use to replicate between hosts, basically each host replicates to the alternative host and then does nightly backups. Each host is in a different closet and each server has dual power supplies, and ups. For critical services like domain controllers, dhcp, dns etc, each host has a running server so most services are redundant if one closest goes down. We have one file share that is not duplicated and that is something that I would want to be able to bring up on the other host if needed. Both hosts are running raid1 for the OS (ESXi) and raid one for the storage, currently hardware raid but that can be changed to jbod if needed, but basically each host has local storage.
I have proxmox running in my lab and seems straight forward for the most part except for the cluster side of things. I understand that you need at least 3 nodes for a quorum or 2 hosts plus a qdevice. I already have a vm hosted on my nas configured as a qdevice but here is my main question. I do not need HA capabilities, again most of my services are already redundant but I would prefer to have a single pane of glass to manage both hosts. From reading I understand that you do not want 2 devices to go down but not sure if that is an issue if you are not running HA for the cluster. My concern and this is what I would like to know is that there are failure situations where I would lose both the qdevice and one of the hosts. Both devices are in the same rack and are served by they same power and switch, if I loose that one switch both the host and qdevice would be down. I know its unlikely to happen but we have had lightening strikes take out network switches before so I do know that while rare it can happen.
What I would like to know is if this happens would I be able to start a VM up on the remaining host if the other host and qdevice are not reachable. I have read mix posting regarding this with some referring to HA and non HA but for me this would be a non HA cluster.
Here is what I am thinking of having, 2 host cluster + 1 qdevice, both running zfs raid1 for OS and another zfs raid1 for storage named the same on both hosts. Both hosts would be added to a cluster along with a standalone qdevice without HA. I would enable zfs replication have the vm's replicate to each opposite host. I think I understand how this would work and understand having the same storage names etc but what I am not sure about is in the rare outage where both a node and qdevice are down would I be able to start up the replicated VM to make it available?
I know I can lab it up but wanted to ask and make sure I understand how it should work in this configuration. The other option would be to run 2 separate nodes with replication and then I just have to remember which vm's are on which host etc, seems like a step back from VMware but just trying to see what my options are.
Thanks
Some background, will try to keep this short, I'm a linux guy for some time, so I understand what is being offered with prox but also looking at xcp-ng and some other options. I originally went with VMware because of cost and the if I get hit by a bus it would be easy for someone to come into and support. Currently we have 2 hosts running various OS, windows server and linux for the most part. Since we have essential we do not have vmotion but we have veeam that we use to replicate between hosts, basically each host replicates to the alternative host and then does nightly backups. Each host is in a different closet and each server has dual power supplies, and ups. For critical services like domain controllers, dhcp, dns etc, each host has a running server so most services are redundant if one closest goes down. We have one file share that is not duplicated and that is something that I would want to be able to bring up on the other host if needed. Both hosts are running raid1 for the OS (ESXi) and raid one for the storage, currently hardware raid but that can be changed to jbod if needed, but basically each host has local storage.
I have proxmox running in my lab and seems straight forward for the most part except for the cluster side of things. I understand that you need at least 3 nodes for a quorum or 2 hosts plus a qdevice. I already have a vm hosted on my nas configured as a qdevice but here is my main question. I do not need HA capabilities, again most of my services are already redundant but I would prefer to have a single pane of glass to manage both hosts. From reading I understand that you do not want 2 devices to go down but not sure if that is an issue if you are not running HA for the cluster. My concern and this is what I would like to know is that there are failure situations where I would lose both the qdevice and one of the hosts. Both devices are in the same rack and are served by they same power and switch, if I loose that one switch both the host and qdevice would be down. I know its unlikely to happen but we have had lightening strikes take out network switches before so I do know that while rare it can happen.
What I would like to know is if this happens would I be able to start a VM up on the remaining host if the other host and qdevice are not reachable. I have read mix posting regarding this with some referring to HA and non HA but for me this would be a non HA cluster.
Here is what I am thinking of having, 2 host cluster + 1 qdevice, both running zfs raid1 for OS and another zfs raid1 for storage named the same on both hosts. Both hosts would be added to a cluster along with a standalone qdevice without HA. I would enable zfs replication have the vm's replicate to each opposite host. I think I understand how this would work and understand having the same storage names etc but what I am not sure about is in the rare outage where both a node and qdevice are down would I be able to start up the replicated VM to make it available?
I know I can lab it up but wanted to ask and make sure I understand how it should work in this configuration. The other option would be to run 2 separate nodes with replication and then I just have to remember which vm's are on which host etc, seems like a step back from VMware but just trying to see what my options are.
Thanks
