# Creating a log volume, using 6 devices, in RAID 10 configuration with 64KB stripes For example, this is an XFS-based example for a log volume. The deployed file system stripe unit ( sunit) and stripe width should match the RAID geometry. Mdadm -create -verbose /dev/md2 -level=raid0 -chunk=64K -raid-devices=2 /dev/sdi /dev/sdjĭisk partitioning and configuration recommendationsįor SQL Server, you should use a RAID configuration. # For tempdb volume, using 2 devices in RAID 0 configuration with 64KB stripes Mdadm -create -verbose /dev/md1 -level=raid10 -chunk=64K -raid-devices=2 /dev/sdg /dev/sdh # For Log volume, using 2 devices in RAID 10 configuration with 64KB stripes Mdadm -create -verbose /dev/md0 -level=raid5 -chunk=8K -raid-devices=4 /dev/sdc /dev/sdd /dev/sde /dev/sdf # For Data volume, using 4 devices, in RAID 5 configuration with 8KB stripes # To locate the devices (for example /dev/sdc) for RAID creation, use the lsblk command In the following example, eight data disks were attached to the Azure Virtual Machine 4 to host data files, 2 for transaction logs, and 2 for tempdb workload. Keep in mind that you should use the appropriate number of data disks for the required throughput and IOPS for volumes based on the data, transaction log, and tempdb I/O requirements. The following example shows how to create software RAID in Linux on Azure Virtual Machines. When configuring SQL Server on Azure virtual machines with similar storage considerations, see Storage configuration for SQL Server VMs. Though, this can differ across different storage vendors and different storage offerings with varying architectures.įor SQL Server on Linux deployed on Azure Virtual Machines, consider using software RAID to ensure appropriate IOPS and throughput requirements are achieved. Normally, in on-premises environments, the storage vendor supports appropriate hardware RAID configuration with striping across multiple disks to ensure appropriate IOPS, throughput, and redundancy. The storage subsystem hosting data, transaction logs, and other associated files (such as checkpoint files for in-memory OLTP) should be capable of managing both average and peak workload gracefully. Storage configuration recommendation Use storage subsystem with appropriate IOPS, throughput, and redundancy Linux OS configurationĬonsider using the following Linux OS configuration settings to experience the best performance for a SQL Server Installation. The following guidelines contain recommendations for configuring both SQL Server and the Linux Operating System (OS). All normal SQL Server recommendations, such as index design, still apply. These recommendations are specific to running on the Linux platform.
This article provides best practices and recommendations to maximize performance for database applications that connect to SQL Server on Linux.
0 kommentar(er)
