I guess I should explain this better. What you are actually doing in your current configuration is mapping a drive from the primary to the secondary and another from the secondary to the primary. The Network Share Data Source is nothing more than a controlled mapped drive from the server that it is attached to. As with all mapped drives, it only points to data in ONE location. To clarify, with this current configuration you have:
Z on the primary machine points to secondary machine\D Z on the secondary machine points to primary machine\D
The purpose of a datasource is to control data that is Shared by all of the servers in the cluster (domain) so configuring the share to actually be on one of the servers themselves would be pointless.
I have some questions for you:
-How far away did the server get moved? -How fast is the pipe between the two servers? -What kind of data is it that needs to be shared? (i.e. is it application data or file share type data)
I've actually had clients with 100 Mb links to servers several miles away and the mirroring Datasource still worked for them. I wouldn't recommend it and I don't think it is supported but I have seen it. Obviously latency plays a big part of this.
My main concern for you is that if the Secondary server was "migrated" to a location that is far away or has a slow link and you are trying to setup a DR type solution here, you will be REQUIRED to invest in some kind of replication software. The network share DS will only let you access data on ONE location NOT two and does not satisfy the requirements for what you are after.
Let me know what kind of data we are talking about here and I'd be happy to help you out with some suggestions.
If the distances between the two nodes is excessive or the network links between them "slow" then consider using RepliStor as the datasource rather than the Windows Mirror Driver. The Mirror Driver uses a synchronous approach to replication whereas RepliStor allows for async data replication.
However, using RepliStor/async replication has its own issues with regards to data in the event of a failover - basically you need to consider what type of data loss can be tolerated in the event of a failover.
Ryan9
45 Posts
0
August 30th, 2007 08:00
Really, the share should be setup on a third server then point both DS to use that Share
Ryan9
45 Posts
0
August 30th, 2007 10:00
Z on the primary machine points to secondary machine\D
Z on the secondary machine points to primary machine\D
The purpose of a datasource is to control data that is Shared by all of the servers in the cluster (domain) so configuring the share to actually be on one of the servers themselves would be pointless.
I have some questions for you:
-How far away did the server get moved?
-How fast is the pipe between the two servers?
-What kind of data is it that needs to be shared? (i.e. is it application data or file share type data)
I've actually had clients with 100 Mb links to servers several miles away and the mirroring Datasource still worked for them. I wouldn't recommend it and I don't think it is supported but I have seen it. Obviously latency plays a big part of this.
My main concern for you is that if the Secondary server was "migrated" to a location that is far away or has a slow link and you are trying to setup a DR type solution here, you will be REQUIRED to invest in some kind of replication software. The network share DS will only let you access data on ONE location NOT two and does not satisfy the requirements for what you are after.
Let me know what kind of data we are talking about here and I'd be happy to help you out with some suggestions.
IKP1
198 Posts
0
September 10th, 2007 09:00
However, using RepliStor/async replication has its own issues with regards to data in the event of a failover - basically you need to consider what type of data loss can be tolerated in the event of a failover.
IKP