May 21, 2015 at 3:41 am
Hello all,
I am testing a SQL Server 2012 AllwaysOn Wsfc implementation.
The architecture i am trying to implement is as follows:
3 Servers. 2 in primary site and 1 in a secondary site.
They will all be part of the Windows Cluster.
Will have automatic failover between clusters.
Manual failover to secondary site.
So Node1 and Node2 will be on primary and Node3 on secondary.
My doubts start at the implementation of AllwaysOn. I want to setup AllwaysOn replication in a dedicated network.
My problem (Just want to clarify how i can make this work) is that in the primary site i will have databases running on both servers. When i implement the failover i want to force the replication to go trough a specific network. Till here its a simple setup.
But when one of the databases that are running on Node1 (and using nic2 of Node1) fails over to Node2 the nic it was using wont exist on Node2. So it will stop replication data to Node3 site...
Can anyone please clarify me on how its the best way to setup this?
Any documentation, etc...
If i wasn't clear enough on my explanation please ask for more details.
Regards,
Arestas
May 22, 2015 at 8:52 am
Arestas (5/21/2015)
I am testing a SQL Server 2012 AllwaysOn Wsfc implementation.The architecture i am trying to implement is as follows:
3 Servers. 2 in primary site and 1 in a secondary site.
This is quite usual.
One question, will the 2 nodes on the primary site host separate standalone instances of SQL server or do you plan to implement a failover cluster instance of SQL Server?
Arestas (5/21/2015)
They will all be part of the Windows Cluster.
You don't have a choice, this is a pre requisite 😉
Arestas (5/21/2015)
Will have automatic failover between clusters.
See my question above, will you be using a failover cluster instance of SQL server on nodes 1 and 2?
Arestas (5/21/2015)
Manual failover to secondary site.
If you use an FCI on nodes 1 and 2 on the primary site you don't have a choice, this is a restriction that is applied when incorporating an FCI as a replica into an AlwaysOn availability group.
Arestas (5/21/2015)
I want to setup AllwaysOn replication in a dedicated network.My problem (Just want to clarify how i can make this work) is that in the primary site i will have databases running on both servers. When i implement the failover i want to force the replication to go trough a specific network. Till here its a simple setup.
But when one of the databases that are running on Node1 (and using nic2 of Node1) fails over to Node2 the nic it was using wont exist on Node2. So it will stop replication data to Node3 site...
Can anyone please clarify me on how its the best way to setup this?
Any documentation, etc...
If i wasn't clear enough on my explanation please ask for more details.
Regards,
Arestas
To offload your mirror traffic down a separate network you'll need to do quite a bit of planning and testing, as well as architecting the necessary network layer(s).
What do you mean by "But when one of the databases that are running on Node1 (and using nic2 of Node1) fails over to Node2 the nic it was using wont exist on Node2. ", why do you assume the network will not exist?
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
May 22, 2015 at 11:26 am
Hello,
Tanks for your awnser.
Yes the 2 nodes on the primary part of a failover cluster instance.
Well, i ll have more than 3 instances running on node 1 and 3 instances running on node 2.
At any point in time (due to server failure or for any other reason) all the instances running on node1 can failover to node2 (and vice-versa).
As an example if you set intance1 that is running on node1 to use the Team ALLON (xxx.xxx.xxx.1) and if after 1 month I failover to that instance to the node2 the Team ALLON will exist (but with a different IP).
Being that I have to set the endpoints by-name or by-ip none of them will exist on node 2…
And that’s my problem….
From what ive been reading I think the solution can pass by configuring a new role on each availability group.
That role would have a Virtual ip on the same network as the Team ALLON and I could move the instances without any problem between node1 and node2.
Just not sure what would happen if I had to manually failover to node 3…
Extra:
On each node I ll have 2 Teams.
Team PUB (public)
Team ALLON (allwayson)
TEAM AO
Node1xxx.xxx.xxx.1
Node2xxx.xxx.xxx.2
Node3xxx.xxx.xxx.3
Tanks for your help,
Regards,
Arestas
May 28, 2015 at 10:01 am
the network teams have nothing to do with it.
Also, each mirror endpoint on each instance is relevant to that instance only, the configurations are held in the AlwaysOn group config.
You're getting yourself confused, have you read my stairway series to alwayson?
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
May 28, 2015 at 10:14 am
First of all i wanna say tanks for the time you ve spend answering my questions.
Because i was with many doubts i decided to try the implementation on my own lab.
And as you said i was confused.
Reading your "stairway to allwayson" (didn't notice you were the author) and some other forums i was able to understand it and to get it working.
Once again tank you very much for your time.
Regards,
Arestas
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply