This site uses cookies to provide enhanced features and functionality. By using the site, you are consenting to this. Read more about our cookie policy

White Paper

Database RAM Performance Summary

As your business grows, your computing hardware often cannot keep up. Decreased performance is not always caused by limited processing power. Sometimes the memory capacity of a server can’t handle the increasing demands of a company’s success. Adding or replacing your servers’ RAM is a cost-effective way to significantly boost performance, while also improving their efficiency—your servers can handle more VMs and distribute their load more effectively. Installing more RAM in your servers can also be far more cost-effective than increasing the number of drives and arrays in a SAN to increase database performance.

Kingston® server RAM is designed to provide you with these benefits. In their labs, Principled Technologies (PT) tested the virtualised database performance of a server with three different memory configurations to show how much improvement you could see in your enterprise environment as you increase memory:

  • 192GB at 1066 MHz = 24 x 8GB PC3-12800R modules

  • 256GB at 1600 MHz = 16 x 16GB PC3-12800R modules

  • 768GB at 1066 MHz = 24 x 32GB PC3-10600R modules (LRDIMM)

The test server ran VMware vSphere ESXi 5.1 and the VMs ran Microsoft Windows Server® 2012 Datacenter Edition with Microsoft SQL Server® 2012. PT used the open-source DS2 benchmark, which provides a workload representative of a real-world database application. Each VM, with 1 vCPU and 25 GB of dedicated RAM, ran a large online transaction processing (OLTP) workload.

PT found that a server with Kingston memory supported the following numbers of test VMs, all while delivering excellent database performance:

  • 192GB of RAM supported 7 test VMs

  • 256GB of RAM supported 10 test VMs

  • 768GB of RAM supported 30 test VMs

PT also measured power utilisation for the three memory configurations. They found that as the amount of memory and number of supported VMs in the server increased, so did the per-workload power efficiency. While running the test workloads, the 192GB configuration used 34.80 Watts (W) per VM, the 256GB configuration used 26.63W per VM, and the 768GB configuration used 13.33W per VM. This makes the 768GB configuration supporting 30 test VMs 62 percent more power efficient than the 192GB configuration supporting 7 VMs.

These results demonstrate the significant improvements in performance that a RAM upgrade can deliver, and show that one of the first considerations for a datacentre upgrade should be increasing the memory capacity of your servers.

        Back To Top