One of the more exciting trends in IT today is the explosion of the hyper-converged infrastructure (HCI) based datacenter. The amalgamation of compute, storage, and networking into a single, software defined infrastructure has really revolutionized the industry. While many consider HCI as strictly an on-premises solution – and many have used it solely for this purpose – it has also become the on-premises component of many hybrid-cloud solutions partnered with various cloud providers. For those of you that have built your datacenters on Microsoft’s Windows® technologies, their Azure® service is also the best candidate for your cloud business. To better enable you, Microsoft created the Windows Server Software Defined Datacenter (WSSD) standard that is built upon their Windows server platform configured with Hyper-V virtualization and Storage Spaces Direct scale-out storage solution.
What’s great about Storage Spaces Direct is that it is designed to take advantage of SSDs (and based on publicly available information on the web, hopefully soon to have support for persistent memory – NVDIMM – in the next release) in both an all-flash or hybrid (SSD cache with HDD capacity roles) solutions. Since Micron is a major player in the SSD and memory space, we are very excited about this and have spent the last several months building some reference architectures (RAs) to help you determine how best to use SSDs in all-flash configurations with and without a cache.
I am happy to announce that we have just published two new RAs that focus on hyper-converged Microsoft Windows infrastructure. The first RA is focused on the value conscious buyer that doesn’t want to compromise on performance. This RA is built as a non-cached, all-flash, SATA solution built upon Micron’s enterprise-class SATA SSDs – the Micron 5100. The second RA introduces an NVMe® cache tier to the same basic configuration using Micron’s enterprise-class NVMe SSDs – the Micron 9200 SSD. Each RA describes building and optimizing a 4-node HCI cluster for hosting virtual machines.
The performance results of for the two different configurations show excellent performance across a variety of I/O profiles and queue depths. The chart below shows just a sample of the performance for both RAs. Our findings show that the introduction of a cache layer using high-endurance, NVMe dramatically reduces the QoS latency of the solution while providing the same basic I/O performance (IOPS). You can see the complete performance analysis, along with a detailed set of configuration information in the respective RAs that are now available.
To read about our Hyper-V/Storage Spaces Direct HCI solutions and download these RAs you can visit our Micron Accelerated Solutions for Microsoft webpage. You can also learn more about our other solutions by visiting our Micron Accelerated Solutions homepage. If you think that this solution can benefit your business, be sure to contact your hardware provider and tell them that you want Micron SSDs in your next Windows HCI server.