We have two data center sites that have a dedicated inter-connection. We also have several legacy boxes that have been in production for 5-10 years (a few with 3-4 year up-times). We need to get away from our physical dependencies while enabling us to provide a greater breadth of services.
Fortunately, a lot of this is going to be new to me; (as I'm a Jr Engineer) I only have knowledge of DRBD, HA Clustering, PXE booting from iSCSI targets and ideal partitioning schemas. This will be fun. ;)
Nonetheless, what we would like to accomplish is to have a HA Cluster at each site that is concurrently running KVM to host our Guest VMs. Make sense? I hope so…
From what I understand, using OpenQRM, DRBD and using iSCSI to boot from the local SAN target, replicated at both sites, we should be able to have our servers fully redundant across both sites.
So, specifically, we would have a controller and "x" nodes at both locations. They would be making use of our DIA in order to synch data / memory usage. The nodes will be scaled as needed in order to provide the necessary resources for additional Guest instances.
One of our Sr. Engineers (with aeons of more experience than me) said that this is impossible and now I'm really curious how I misunderstood this. Is this possible, and if not, where have I strayed? :)
Once we get that in place, I'd like to create redundant clusters, so if a controller fails and that entire node goes down, the node at our other site would bare the full user-load while the second, redundant cluster in the first, primary site boots up and comes online. Is this possible or purely "in an ideal world"?
Has anyone implemented similar infrastructure? I'd LOVE and TRULY APPRECIATE your insight.
I have a lot of trial and error that will need to take place before I even get ready to present this to the rest of the IT department, but I want to make sure that I come to them with something "nearly production ready". So, with that being said, I am all ears and look forward to any advice and wisdom that you all may have to share.
Thanks in advance.
Sincerely,
Dro Buddy
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
sure this is possible to setup with openQRM. I would like to recommend to contact openQRM Enterprise. They are experienced in such kinds of custom DR setups.
many thanks and have a nice day,
Matt Rechenburg
Project Manager openQRM
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
We have two data center sites that have a dedicated inter-connection. We also have several legacy boxes that have been in production for 5-10 years (a few with 3-4 year up-times). We need to get away from our physical dependencies while enabling us to provide a greater breadth of services.
Fortunately, a lot of this is going to be new to me; (as I'm a Jr Engineer) I only have knowledge of DRBD, HA Clustering, PXE booting from iSCSI targets and ideal partitioning schemas. This will be fun. ;)
Nonetheless, what we would like to accomplish is to have a HA Cluster at each site that is concurrently running KVM to host our Guest VMs. Make sense? I hope so…
From what I understand, using OpenQRM, DRBD and using iSCSI to boot from the local SAN target, replicated at both sites, we should be able to have our servers fully redundant across both sites.
So, specifically, we would have a controller and "x" nodes at both locations. They would be making use of our DIA in order to synch data / memory usage. The nodes will be scaled as needed in order to provide the necessary resources for additional Guest instances.
One of our Sr. Engineers (with aeons of more experience than me) said that this is impossible and now I'm really curious how I misunderstood this. Is this possible, and if not, where have I strayed? :)
Once we get that in place, I'd like to create redundant clusters, so if a controller fails and that entire node goes down, the node at our other site would bare the full user-load while the second, redundant cluster in the first, primary site boots up and comes online. Is this possible or purely "in an ideal world"?
Has anyone implemented similar infrastructure? I'd LOVE and TRULY APPRECIATE your insight.
I have a lot of trial and error that will need to take place before I even get ready to present this to the rest of the IT department, but I want to make sure that I come to them with something "nearly production ready". So, with that being said, I am all ears and look forward to any advice and wisdom that you all may have to share.
Thanks in advance.
Sincerely,
Dro Buddy
Dear Dro,
sure this is possible to setup with openQRM. I would like to recommend to contact openQRM Enterprise. They are experienced in such kinds of custom DR setups.
many thanks and have a nice day,
Matt Rechenburg
Project Manager openQRM