What extra work or differences to backup a cluster than back up regular servers ? If there are three nodes in it, only one node or shared file systems on it will be available, when I back up the cluster, should I just back up the active server or file systems? Do I need to back up those not active nodes as well? What issues it would cause if I just back up three respective nodes as the way for backing up the luster?
As you can tell now, I am seeking for a steps how we should back up a Window Cluster? I am new to NB backup, and I am more like a Storage admin.
In addition, there are several shared data LUNs assigned to each server. There is also a lun called "quorum" for each cluster, only one "witness " lun for one of clusters.
What is this "quorum" for?
and What is "witness" lun for?
Thank you for your inputs!
after looked the link, still don't know the answers of my questions above. Can you please elaborate my questions in a high level please?
Why do we have to back up each node and shared disks, and as well as virtual server in different scheduled times?
What is the virtual server referring to? Is this a cluster node versus each individual node
"quorum" and "witness" are microsoft terms and functions, so you must ask a microsoft expert.
In order to backup a cluster you have to create one policy for the nodes and one policy of every virtual hostname (service) you have in the cluster.
With the first policy you have to backup the local drives of every node using the node name as client (and backup the registry also)
Then you have to backup the cluster recurses with a new policy using the service hostname as client. (shared drives if the cluster is a fileserver or SQL backup if it is SQL service)
Yes, that's what I am looking for.
In the second policy, I specified service hostname as client (I guess this is the name for a node from cluster perspective), what details would be included and be backed up, other than shared drives?
When you said FS, did you mean File Server?
Another silly question, can you give me 1 or 2 examples as for consequences what if I didn't perform the backup based on 2nd policy? I know this backup is to back up the cluster, but all drives on all nodes should already been backed up in 1st policy.
Also, what is the point to schedule backups for nodes and cluster (2nd policy) in a different time?
Policy #1 - backs up each cluster node using ALL_LOCAL_DRIVES directive. Should have an exclusion for any cluster resource drives that might be mounted during the backup. This policy ensures that you have all of your configuration data, your OS data, local user desktops, registry entries, etc. Everything that you might care about that does not fail around with the cluster drives.
Policy #2-100 - One policy per hostname associated with the cluster resources. This way it does not matter what node the cluster resource happened to be living on when the backup kicked off, you always will just restore from the same hostname; otherwise you'd have to play guessing games in the middle of a critical restore and that's never any fun. This also allows you to easily split up the backup load on the cluster into different windows to decrease user impact, plus potentially allowing you to tailor the filelist to be specific to the directory layout on the drive. For example, instead of having just an H:\ stream you could run it as H:\* and get multiple simultaneous backup streams, potentially increasing your aggregate throughput rate (depending on the HW).
Hope that helps.