Cloud Insights from Brad Anderson, Corporate Vice President, Windows Server & System Center
I’ve heard IT departments all over the world share a very similar joke about how the rest of the company has no idea where their e-mail comes from or where their SharePoint sites are stored – they just want these things to be there when they need them.
IT departments, however, do know exactly where these e-mails come from and where the SharePoint data resides, and they also know that these apps need to be scalable, high performance, and provide enterprise features and capabilities right out of the box. This fact is something we focus on constantly, and teams throughout Microsoft have spent millions of hours (22.4 million on System Center alone!) building platforms – like Windows Server, Hyper-V, System Center, and SQL Server – that are fine tuned for these workloads. Simply put: Microsoft platforms get the most out of workhorse applications like SQL, SharePoint, and Exchange.
Running Microsoft workloads on non-Microsoft virtualization solutions is a lot like building a house with stone from two different quarries. At the end of the day, you may simply be glad to have a house, but it can be a big gamble to use two unrelated kinds of building material on something so important. If nothing else, you run the risk of your house looking a bit like the Washington Monument.
Without a world-class infrastructure, many users of SQL, SharePoint, Exchange, and other workloads may never see the full spectrum of what these applications can really offer in terms of scalability, compute power, and overall performance.
I’d like to examine each of these three metrics in this post.
When we think about scalability and compute power, Hyper-V has emerged as a particularly dynamic product. By the numbers, Hyper-V represents the single best option for customers looking to virtualize their mission critical, tier-1 applications and workloads at the lowest cost possible.
Consider this: Hyper-V delivers 64 vCPUs and 1TB RAM per VM, with no vSphere-like SKU specific restrictions, and, on top of this, Hyper-V supports double the physical host size of vSphere 5.1. And, on top of all that, if an IT team needs to dramatically scale out their resilient, highly available infrastructure, Hyper-V supports up to 64 physical nodes, and 8,000 VMs per cluster, double that of vSphere 5.1. And this functionality is all available right out of the box.
If you’d like to get deeper into these figures (and I highly recommend it), take the time to read this highly detailed whitepaper.
One graphic I find particularly interesting within the document is this simple comparison:
With these scalability and compute figures in mind, let’s look at the performance metrics from 3rd party testing of Microsoft workloads like SQL and Exchange on our management software.
The performance of Microsoft workloads on Microsoft management technologies is something that has been carefully and exhaustively measured, and the results are pretty surprising.
The 3rd party testing of SQL Server 2012 running on Windows Server 2012 Hyper-V examined an existing SQL Server 2012 OLTP workload that was previously vCPU limited and, in their words, “increased the performance by six times, while the average transaction response times improved by five times.” The tests also discovered that, Hyper-V’s overhead of 6.3% was recorded “when comparing SQL Server 2012 OLTP workload performance of a physical server to a virtual machine configured with the same number of virtual CPU cores and the same amount of RAM.”
(To read a detailed technical analysis of this testing, you can view the full report here. I also recommend the SQL+Windows Server+System Center “better together” overview, and the SQL+Hyper-V best practices documents for additional information.)
Also consider how System Center 2012 SP1 can streamline the deployment and management of SQL Server in a virtualized environment. The benefits include the ability to rapidly deploy new standardized SQL Server virtual machines, support for self-provisioning of new SQL Server virtual machines, the creation of an IaaS platform that can request and provision new clouds or VMs, improved data security, and reduced downtime.
Windows Server 2012’s support for Exchange workloads is another area that was closely examined. In a test that deployed Exchange 2013 on twelve Hyper-V virtual machines running on a single physical server with 48,000 simulated users, we saw these results: “Average database read response times ranged between 5.02 and 15.31 milliseconds, well below the Microsoft recommended limit of 20 milliseconds.”
(To get into the details of these tests, check out the full report here – and also check out the best practices whitepapers about virtualizing Exchange 2013 and Exchange 2010.)
When System Center 2012 SP1 is used to manage Exchange in a virtualized environment, users can more effectively monitor, maintain, and protect Exchange data. These upgrades are accomplished by enabling IT administrators to deploy Windows Server VM’s to host Exchange workloads, creating role-based service access for administrators, as well as the standard System Center support for IaaS, performance insight, data protection, and reduced downtime.
One last note: Testing has just been completed on a single Hyper-V host, running 5 SharePoint virtual machines, scaled to over 1 million users. This test showcases, in great detail, Hyper-V’s ability to drive the highest levels of performance for enterprise workloads. The report on this test is currently being written.
To see what is possible with SharePoint workloads, check out the solution I introduced during my keynote at MMS – the Service Template for SharePoint 2013. This service template has generated a lot of interest about what can be done with an on-premise private cloud – and you can download the service template here. What makes this offering interesting to me is not what it is doing, but how it relates directly back to your own strategies for workload deployments in your Microsoft private cloud. This solution is also usable by hosters who want a process that is repeatable and streamlined for customer use.
With these workloads running at optimal speeds, with optimal compute power, and with optimized performance – we are offering the industry’s most sophisticated tools for managing apps, on-prem and cloud-based infrastructure, as well as PC’s and devices. With Windows, Server, Hyper-V, Windows Azure, and System Center, IT departments with rapidly growing infrastructures and workloads can execute some very important operations: Accelerate deployments, draw out complex insight from their operations, and centrally protect, automate, and efficiently manage key applications and workloads running on the Microsoft Cloud OS platform.
By running Microsoft workloads and platforms, IT administrators are benefiting from the technology and tools that have been meticulously fine-tuned to work together.
Are there any white papers contrasting this workload on vSphere 5.1 Enterprise Plus versus Hyper-V?