Virtual disk files.You can create virtual disks for VMs from existing datastores and provision SQL Server databases on them. This provides the highest level of flexibility by managing storage provisioning from the virtualization layer, so resources can be used most effectively. Microsoft and NetApp recommend that you host applications such as SQL Server databases on dedicated LUNs even if VMDKs are used in the guest. This configuration permits granular control for backup, recovery, and replication purposes and allows you to manage storage performance on a per-database basis. Raw device mapping.Customers can also provision LUNs for SQL Server databases that are allocated to the ESXi servers but then passed directly to the VM as a raw device. This arrangement has many of the same benefits as VMDKs, but it imposes serious limitations in the operational capabilities of the vSphere environment. RDM devices are becoming the last choice in deployments due to these limitations and the administrative overhead created by using RDM disks. This solution is based on VMDK virtual disks, and as seen in the section “Solution Verification,” this configuration delivers ample performance for the SQL Server workload. All database files should be on separate virtual disks from the OS and separate from each other. Database data and log files should be on separate dedicated LUNs, and tempdb should be on a separate LUN as well. LUNs should be mounted to servers by using volume mount points, which creates a more logical data structure and prevents you from running out of drive letters on the host. SQL Server AlwaysOn Availability Groups AlwaysOn Availability Groups are an enterprise-level high-availability and disaster recovery solution introduced in SQL Server 2012 that enable you to maximize availability for one or more user databases. The use of AlwaysOn Availability Groups requires that SQL Server instances reside on Windows Server failover clustering (WSFC) nodes. For this validation, a single WSFC was created with 12 nodes. Six availability groups were created with two nodes each, and each group supported a single database. The availability groups were configured in synchronous commit mode, providing the highest level of data protection. When you host the primary and secondary copies on high-performance hardware with local network connectivity, the performance effect of synchronous commit mode is minimal and offers completely redundant and high-performing replicas of mission-critical databases. 6 Solution Verification This reference architecture is based on a standard FlexPod infrastructure hosting Microsoft SQL Server 2014 databases. Cisco Nexus 9296PX switches in NX-OS mode connect the Cisco UCS compute nodes and NetApp storage arrays to the Ethernet network for client and NAS storage access. The storage arrays are connected directly to the Cisco UCS fabric interconnects for FCoE storage access. Cisco UCS B200-M4 blades running VMware vSphere ESXi 6.0 host Microsoft Windows Server 2012 VMs and use dedicated FCoE LUNs for individual database instances. Specific details of the configuration can be
You've reached the end of your free preview.
Want to read all 56 pages?
- Spring '17
- ........., Microsoft SQL Server, NetApp