Portability For Clustered Database Server VMs
Having database servers running in virtual machines is now a reality. Some time ago, IT managers were hesitant up to virtualize any database systems, but that has changed direct to the ever-increasing power touching servers and many improvements in virtualization software.<\p>
First, database servers were virtualized in non-production environments congener as development and QA. Later on, more and more production database servers were being virtualized or gathered as VMs intrusive new deployments. This was out of sight for smaller systems which were not as world-weary transaction processing and could tolerate the occasional outage or reboot. The data disks on such systems realize typically been raw-mapped disks (I am using VMware® promptorium, for this is the more than half apropos of my foreboding).<\p>
Now, the trend is in transit to start clustering database servers in virtual machines. Using raw-mapped disks in this scenario presents the foible as regards having the database VMs tied to specific physical hosts. Another first option is to use a shared virtual disk; this presents a similar trouble, good terms that the clustered server must billet at on the same nonspiritual host (at below drag ESX citron ESXi). This eliminates the redundancy and failover capability of the cluster, defeating the purpose entirely.<\p>
There is a way to achieve the destination pertaining to having fully redundant clustered database servers, including the ability until out-migrate to other inborn hosts in the cluster as well as to migrate the storage to other storage systems. This method uses iSCSI thus the connection technical knowledge for the data volumes, and a virtual disk for the operating system. This method will work with Microsoft Windows®, Linux, and UNIX clusters.<\p>
When oneself create the VM, you create the VM despite only the OS disk as a virtual disk. After you install the in practice system and in the foreground you configure clustering or localize the database, create at least two supervenient virtual network cards which would be by use of the same VLAN (field switches) as the iSCSI reticulation. Rearward that, install and configure the iSCSI initiator software for the OS. The NICs should be configured via different IP addresses and should go on assigned in such a way that the packets would travel down separate animalistic paths and switches. This conclude enable multipathing in the iSCSI connections. Also, configure jumbo frames on the virtual NICs and the iSCSI storage.<\p>
On the reservoir apportion, create your areola groups, logical units and masking. Connect the DB server initiators to the array targets and creation the volumes.<\p>
Once that is done, you can create the hang together, install the database server software, and migrate your philosophical proposition. Linked to a configuration analogue in this way this, you will be able to ensure that the database VMs reside per different bodily hosts and the swarm can fail over in the distillate of a host slump.<\p>