Languages Business Resources Where To Buy
Memory
SSD
USB Drives
Flash Cards
Wireless
Support
MEMORY SEARCH

Memory and SSD Optimization

Optimizing Memory and SSD in Servers

If your IT department is considering a roll-out of Windows Server 2012 and SQL Server 2012, it is important to have an understanding of server memory behaviour and the impact of solid-state storage on key release features of both platforms. In a paper authored on behalf of Kingston by VDX, a premier consulting company specializing in creating flexible desktop and data centre solutions, a deep-dive was conducted into virtualization and SQL Server workloads, two areas that represent the best places to consider optimization of both server memory and SSD to fully utilize new capabilities and performance improvements available in the releases.

Although the paper goes into full detail, the following basic guidelines highlight server memory optimisation and best practices for using of SSDs to tackle high I/O challenges for the Windows 2012 version of Hyper-V and for designing the best back-end hardware architecture for SQL Server 2012 use in the data centre.

Memory and SSD optimisation best practice considerations and recommendations:

  • When provisioning the physical server, always ensure that memory is populated properly in the memory banks following triple-channel or quad-channel memory architecture and NUMA. Visit http://www.kingston.com/en/business/server_solutions for more information on optimising physical server DRAM configurations for performance, capacity and/or power.
  • When leveraging Hyper-V, use Dynamic Memory to ensure the investment made in server memory is used most effectively. Take extra care with workloads that may have memory configuration settings which are not configured optimally, such as database applications – these could lead to unpredictable results with Dynamic Memory use. Always understand the true memory and storage performance requirements of any workload to be virtualised.
  • Consider placing guest paging files on separate VHDs housed on SSDs, particularly for critical workloads. This will ensure that if the Hyper-V host memory becomes constrained, the resulting in-guest paging will provide the best possible performance. You should also consider using SSDs for Hyper-V Smart Paging files to optimise performance in memory-constrained situations.

 

  • When virtualising SQL, do not oversubscribe CPU and memory on the server. Analyse the SQL Server instance(s) to determine the memory requirements and then configure the max server memory setting to prevent unnecessary consumption of memory for the buffer pool. Hyper-V Dynamic Memory may or may not benefit SQL Server. If the instances are properly configured for max server memory, then the amount of memory required for the VM is predictable and Dynamic Memory may not be required. If the max server memory setting will not be configured, use static memory in the VM configuration; do not use Dynamic Memory and only place one SQL instance on a server.
  • Use server-grade Enterprise SSDs for specific SQL databases or transaction logs. The SQL Tempdb will benefit from the SSDs’ performance and can easily be leveraged in a local server to avoid steep SAN costs. Rather than placing each database’s transaction logs on a different storage volume, consolidate them onto a single SSD array to benefit from the increased performance. To learn more about the Kingston Enterprise SSD offerings, please visit http://www.kingston.com/en/ssd/enterprise.
  • Non-persistent VDI storage performance will benefit significantly from placing differencing (delta) disks on SSD storage. And, although challenging, non-persistent VDI is the target model for a successful ROI on VDI solutions.

For the complete white paper on server DRAM and SSD optimisation in Windows Server 2012 and SQL Server 2012 and to learn more about best practice recommendations and optimisations to consider when determining the right strategy for your IT environment, read the Memory and SSD Optimisation in Windows Server 2012 and SQL Server 2012.