Sorry if this should be in the Scale forum instead, from what I understood the clustering part is going to be more a feature of Command so asking here.
In general I'm just trying to learn more about the high availability/load balancing options that are going to be available. I know right now creating 'Cluster Volumes' really is just making glusterfs volumes available to mount from other servers, which is perfectly fine for our linux servers that just need access to some sort of network storage.
Say we have 3 servers available to cluster, my main questions regarding SMB, NFS, and iSCSI are
1. I know SMB is on roadmap for 21.08, any plans for NFS/iSCSI?
2. Are the shares/targets going to live "inside" of the glusterfs volumes so that glusterfs is handling the replication, etc? Is it known yet what sort of impact that would have particularly on iSCSI?
3. Will we be able to point to all 3 servers simultaneously to load balance demand?
4. If yes to #3, if any 1 server goes down , is there a mechanism to automatically failover to one of the others so that the connection doesn't fail completely, like how glusterfs works?
5. If no to #3, is the goal more of a failover scenario, where we need a virtual IP point at a system that all services are running on properly?
Hope this is clear and thanks for any information. Anything else you can add to how specifically it's planned to work, like if NFS will rely on ganesha support in glusterfs would be appreciated.
In general I'm just trying to learn more about the high availability/load balancing options that are going to be available. I know right now creating 'Cluster Volumes' really is just making glusterfs volumes available to mount from other servers, which is perfectly fine for our linux servers that just need access to some sort of network storage.
Say we have 3 servers available to cluster, my main questions regarding SMB, NFS, and iSCSI are
1. I know SMB is on roadmap for 21.08, any plans for NFS/iSCSI?
2. Are the shares/targets going to live "inside" of the glusterfs volumes so that glusterfs is handling the replication, etc? Is it known yet what sort of impact that would have particularly on iSCSI?
3. Will we be able to point to all 3 servers simultaneously to load balance demand?
4. If yes to #3, if any 1 server goes down , is there a mechanism to automatically failover to one of the others so that the connection doesn't fail completely, like how glusterfs works?
5. If no to #3, is the goal more of a failover scenario, where we need a virtual IP point at a system that all services are running on properly?
Hope this is clear and thanks for any information. Anything else you can add to how specifically it's planned to work, like if NFS will rely on ganesha support in glusterfs would be appreciated.