Hi all,
I am new to the concept of clustering, so please enlighten me.
I manage a domain with 5+ RD session host servers and 10+ misc admin/support servers (DC, file servers, webservers, RD Gateway etc). All servers are running windows server 2008 R2, domain functional level at 2008 R2. RD session hosts serve heavy
computational tasks, very little web/network activities (mainly through RD Gateway for RD traffic).
Now I am considering to reorganize the domain to provide high availability and load balancing for RD session hosts.A planned investment of 2 high-powered servers with a central storage device (Direct Attached Storage with mixed SSD and SAS disks) is considered.
I am inclined to deploy VDI remote desktop solutions to 30+ users.
I am considering a 2-node clustering, with a common DAS storage. From what I have read, in a failover clustering scenario, one node is idle/passive and another node is active in normal operation. Should one node fails, another kicks in and all VMs would be moved to the active node. This seems a lot of wasted hardware. I have less pressure for downtime (not required to have operational status 24/7), but more for computational power.
So I am interested in mostly a load-balancing cluster with failover feature. My questions are these:
1. Can I achieve the goal that VDI vms are evenly distributed to the two nodes in normal operation? At the same time when one node fails, all VMs can be moved to the other nodes without down time?
2. Is it correctly understood to say that the clustering is archived at physical server level, that is, two identical physical servers are configured in cluster mode, with access to the storage device as Cluster Sharing Volume. On top of each physical servers, hyper V visualization hosts are installed and provision the vms. While load balancing is achieved by RD connection broker to distribute VMs across two visualization hosts.
3. so my ideal configuration would be like this (from bottom to top):
Node1 || Node2
physical server1 || physical server2
hyper V host1 || hyper V host2
RD connection broker1 || RD connection broker2
VM1-VM10 || VM11-VM20
In case of failure of say Node1, all VMs will be moved to Node 2 for continuous operation.
Can this configuration be achieved? Please correct me if I have misunderstood the concepts. What requirements are there to achieve this?
Thank you very much.