Enterprise Data Center Infrastructure Hardware Update considerations

"Tenkine Server channel July 8 message" Performance improvement and maximum productivity per http://www.aliyun.com/zixun/aggregation/13999.html > workload is the best way for companies to get more out of their hardware investments. So the enterprise in the infrastructure of the update input is cautious, then how the enterprise by updating hardware to get the due roi?

Performance is often related to the server hardware itself, but does not always refer to the purchase of the latest and most famous brand series. In many cases, potential upgrades or adjustments to existing devices can also bring measurable improvements, such as CPU and storage.

Choosing a faster 64-bit processor is the key. For virtualized scenarios, the processor should also support level two address translation slat to convert physical memory addresses to virtual and directly enhance hypervisor performance. Slat is called an extended page table or EPT, used for Intel chips and nested page tables, or for the NPT of AMD chips.

Processor cache size is also a matter of computational performance. The processor stores recently accessed instructions and data as a relatively small area of fast memory. Check the cache first when obtaining any instruction or data. When the required content is in the cache, the processor does not need to get content from system memory, which greatly enhances performance. Generates a normal memory access period when the desired content is not in the cache. Large-capacity L2 and L3 caching often benefit system performance.

Select the memory module that meets the processor clock speed. Choose to match the processor's best bus speed. The benefits of the amount of memory are similar, ensuring that there is enough memory to support the operating system and all workloads on the system.

Most storage and network interface devices now rely on the PCI Express interface, so be sure to select a server with a sufficient number of PCIE interfaces. If the storage is local (instead of a storage area network live network attached storage), to improve storage performance, you can select a small disk with high RPM, and use multiple disks from a raid group to spread across multiple read and write activities for concurrency. Some read-intensive or latency-sensitive applications can also benefit from solid-state disk devices rather than traditional mechanical disks.