06-02-2015 12:06 AM
hi,
i wish to ask a couple of this all related.
i could have failover and parallel service groups in the same vcs.right?
in my opinion would be better if i setup a different vcs only for failover services and another for parallel services rather that a single vcs that comprises both failover and patrallel service groups.what is your opinion?
in an vcs spread accross 2 sites within same vlan,the failover service group should has the data in sync between sites and if i have an parallel service group with a node in a site and another in other site,than the dg will be made of mirror volumes (a submirror in one site and other in the other site).right?
if vcs start an a certain node and normally an certain service group start on this node,but in this scenario the service group runs on other vcs node,than the new node that joins will not start the service group.right?
tnx a lot.
Solved! Go to Solution.
06-02-2015 12:21 AM
06-02-2015 04:52 AM
in my opinion would be better if i setup a different vcs only for failover services and another for parallel services rather that a single vcs that comprises both failover and patrallel service groups.what is your opinion?
>>> Marianne is right, its common to have clusters with both parallel & failover service groups. for e.g Oracle RAC can run in parallel groups & while clusterservice which is a default service group will run in failover mode.
in an vcs spread accross 2 sites within same vlan,the failover service group should has the data in sync between sites and if i have an parallel service group with a node in a site and another in other site,than the dg will be made of mirror volumes (a submirror in one site and other in the other site).right?
>>> Correct, data has to be in sync across 2 sites. If your sites are far across, You can use VVR to achieve replication & GCO to run global cluster across sites. If your sites are closeby, then mirroring across sites is a common & surely can be used in Parallel SGs.
if vcs start an a certain node and normally an certain service group start on this node,but in this scenario the service group runs on other vcs node,than the new node that joins will not start the service group.right?
>>> When you have a running cluster, LLT & GAB will be seeded with cluster ID & no. of nodes GAB has to register membership. If you simply introduce new node, GAB will not form membership with it. So as rightly referred by Marianne, refer admin guide which has appropriate section of adding node to cluster.
G
06-02-2015 12:21 AM
06-02-2015 04:52 AM
in my opinion would be better if i setup a different vcs only for failover services and another for parallel services rather that a single vcs that comprises both failover and patrallel service groups.what is your opinion?
>>> Marianne is right, its common to have clusters with both parallel & failover service groups. for e.g Oracle RAC can run in parallel groups & while clusterservice which is a default service group will run in failover mode.
in an vcs spread accross 2 sites within same vlan,the failover service group should has the data in sync between sites and if i have an parallel service group with a node in a site and another in other site,than the dg will be made of mirror volumes (a submirror in one site and other in the other site).right?
>>> Correct, data has to be in sync across 2 sites. If your sites are far across, You can use VVR to achieve replication & GCO to run global cluster across sites. If your sites are closeby, then mirroring across sites is a common & surely can be used in Parallel SGs.
if vcs start an a certain node and normally an certain service group start on this node,but in this scenario the service group runs on other vcs node,than the new node that joins will not start the service group.right?
>>> When you have a running cluster, LLT & GAB will be seeded with cluster ID & no. of nodes GAB has to register membership. If you simply introduce new node, GAB will not form membership with it. So as rightly referred by Marianne, refer admin guide which has appropriate section of adding node to cluster.
G