vRanger 7.8.6 - Integration Guide for NetVault SmartDisk

Understanding the optimal performance strategy

Consider the following guidelines when assigning NetVault SmartDisk Storage Pool Roles:

•

Content Index: Should be small and reside in a Storage Pool made of fault-tolerant disks with good random-access performance.

•

Staging Store: Should reside in a Storage Pool made of fault-tolerant disks with good streaming performance.

•

Chunk Index: Should reside in a Storage Pool made of fault-tolerant disks with good random-access performance.

•

Chunk Store: Should reside in a Storage Pool made of fault-tolerant disks.

If the goal is optimal performance, using additional disks to increase Input/Output Operations per Second (IOPS) and aggregate disk bandwidth improves NetVault SmartDisk performance more than using additional memory.

For example, if you are using a single RAID array, consider configuring four separate RAID 1 volumes, one for each Storage Pool Role. This configuration separates the IO workload for each volume across independent RAID volumes. You can tune performance further by restricting the number of independent disk operations — reads, writes, and deletes — allowed per volume to two operations, which avoids disk thrashing. In addition, configure the RAID array to support more bandwidth in and out of Staging Store.

This strategy is appropriate for NetVault SmartDisk Instances where staging, deduplication, and restores might occur simultaneously.


	NOTE: RAID recommendations also apply to Storage Area Network (SAN) and Network Attached Storage (NAS).

On Windows® platforms, you can disable the New Technology File System (NTFS) Last Access Update feature, which can reduce disk accesses and increase performance. For instructions on disabling this feature, see the documentation for your specific operating system.

Optimizing performance while protecting against data loss during power failures

Although enabling disk-write caches improves NetVault SmartDisk performance, power failures that occur before modified disk-cache contents have been written to non-volatile magnetic storage can potentially cause data loss in NetVault SmartDisk. Because of this risk, it is critical that you understand how your underlying disk technology caches writes to disk. You can turn off disk-write caching, but due to the improvement in performance that write-caching offers, it is increasingly used despite the risk, and the risk is mitigated by using additional technology. A common mitigation technique is ensuring that power does not go off. In high-end server environments, with their uninterruptible power supplies (UPSs) and redundant power supplies, having unfilled cached writes is less of an issue.

Also, drives that employ write-caching have a write-flush feature that instructs the drive to send pending writes from the cache to the disk immediately. This command is sent before UPS batteries run out — if the system detects a power interruption — or just before the system is shut down for any other reason.

Finally, most disk array systems use non-volatile random access memory (NVRAM) to protect data written to disk if there is a power failure. Quest recommends that you review your server and storage vendor’s product documentation to understand what steps are taken to ensure that disk writes are written to non-volatile magnetic storage if there is a power failure.