Switching from VMware to Microsoft's Hyper-V virtualization environment is saving money for two businesses that are also using Microsoft's System Center 2012 to manage and monitor applications and the infrastructure that supports them.
Both businesses say they were attracted by the Systems Center bundle of management tools for data centers and client machines that can help with monitoring, configuration and security that gives them features they need now and that they may need in the future. If they do, they won't have to pay more, they say.
VIRTUALIZATION FEUD: VMware official fires back at open source cloud vendors
Both businesses, The Walsh Group construction company and online apartment rental business Apartments.com, say they were driven to the Microsoft offerings by VMware's pricing. And because of Systems Center 2012 features, are both considering whether they can take advantage of cost savings by adopting public cloud services.
The Walsh Group regarded using the Microsoft platform as already paid for as part of its enterprise agreement with the software maker, and started a pilot of Hyper-V as a VMware replacement, says Patrick Wirtz, the firm's manager of technology innovation.
The VMware-based data center had a combined total of about 200 physical and virtual servers, and it supported more than 150 sites with 100 virtual machines initially, plus more at job sites. The 12 regional offices each have a physical server running four virtual machines.
The company uses 3D modeling for its auto CAD applications that are used in remote locations by engineers. It didn't perform that well with VMware, but Microsoft's Remote FX could use the 3D cards on workstations to support the auto CAD, Wirtz says. "There was no such option in VMware," he says.
To start its transition, Walsh Group installed Hyper-V virtual machines on new physical server as the company needed more capacity. Gradually it is replacing VMware with Hyper-V on existing servers as well. Today about 80% of its environment is Microsoft.
Most of the sites left to transition are regional offices with no IT staff, and they will migrate to Hyper-V when the physical servers there need to be replaced.
Staffers needed to study up on Microsoft live motion replication of virtual machines vs. VMware's VMotion, but that was relatively painless because most of the IT group is versed in Microsoft. Older versions of Live Motion were more involved than later versions, so required more education.
The company started using System Center 2012 beta in its development environment in October 2011, using one component of it, Virtual Machine Manager, heavily, then switching System Center into the production network about two weeks ago with the release-to-manufacturer version.
Managing job sites is easier with System Center 2012, Wirtz says, particularly through use of two other components, Orchestration Manager and Operations Manager, which enable configuring virtual machines to pick up the function of others when they go down, he says. Help desk workers can perform this task without escalating for help from IT admins. "The power is in the hands of the people that need to solve the problem," he says.
He says he expects the combination of Server 2012's management plus putting SQL Server 2012 on virtualized servers will result in more efficient use of all the processor cores in the company's physical servers.
The company this year is going to adopt not only System Center and SQL Server 2012 but also Server 2012 as well. "There's a little bit of growth pain there," he says. "It will only help us in the long run to start sooner rather than later."
The company has talent in-house to handle the transitions without too much trouble because they've done it before.
While the company doesn't use private cloud services yet, it is an attractive option for tasks that require extra resources temporarily. They need to consider Amazon and Azure cloud services, but haven't made the leap yet.
He likes the idea of a suite of management tools rather than buying them just as needed because it gives the company options for expanded services without seeking further funding.
Putting the power to actually address virtual machine problems in the hands of the help desk gets end users back to work faster and ties up less time of administrators, who can then work on more strategic projects, Wirtz says.
Over at Apartments.com, Director of Technology Operations Matt Stratton is reaping similar benefits from pushing authority to create virtual machines in the hands of the users who need them.
In particular System Center will save time by enabling developers to create virtual machines themselves to simulate the production environment and in which they can test new applications. Not having to call in IT help and automating steps will cut production time for new applications by 10% to 20%, Stratton says.
The company had been running VMware managed by VCenter for its virtualization, all for running virtual application servers, but not VDI.
The company considered System Center because it has standardized on Windows Server and VMware requires a different skill set. If the company needs to hire new staff, it's easier to get someone with Microsoft expertise than it is to get someone with both Microsoft and VMware expertise, he says.
The company will achieve a 70% virtual machine operating-cost savings by cutting VMware out of the picture, says.
The company funded a change from VMware to Microsoft in 2011 and implemented it during December and January. Now the company runs 20 Hyper-V hosts supporting about 350 virtual machines.
The company is running Virtual Machine Manager 2008 R2, but is looking to upgrade to 2012 during this quarter, initially just maintaining current functionality. But that will lay the groundwork for developing a private cloud later, Stratton says.
The company has two data centers, one of for development and one for production. Hyper-V ran in the test and development center first, where all the physical servers were up for a refresh. They created a Hyper-V cluster there. He says the development data center is still a production environment where the company's customer-facing applications are produced and where quality assurance is run.
The transition to Hyper-V took about three weeks, and involved transitioning from HP Proliant servers to Cisco UCS blade servers. The project was completed in December and work on the production data center started Jan. 1.
The production data center is not getting a hardware upgrade, although some have been added as a buffer for growth and redundancy. Existing machines are being transitioned in a phased approach, Stratton says. As VMware virtual machines are replaced with Hyper-V VMs, they are clustered.
The transition took longer because more extensive testing was involved. If there were problems, they had the option to fire up VMware again. The actual switchovers were done during low-usage times, and it took about three hours per virtual machine.
The production network is mainly Windows Server machines, but also includes about 70 Linux servers running on HP Proliant boxes.
The company is doing a swap-out with no change in functionality initially, but it has proved a good opportunity to find unused VMs that were created and never removed once need for them dropped off. "It's a good opportunity to get rid of a lot of junk," he says.
The company worked with Microsoft to help translate terms used in VMware to those used in Virtual Machine Manager, and staff, already familiar with virtualization concepts, got comfortable with it within a week or two, he says.
The impetus to move from VMware was financial, he says, and initially he didn't see the potential for adding capabilities. But he says the elasticity that Systems Center supports will help streamline operations down the road. More people sign leases in the summer than the winter so scaling the infrastructure using public cloud resources is attractive financially.
Stratton is not leery of public cloud for security reasons but is concerned that resources in public clouds might not be compatible with his corporate infrastructure. The goal is for performance to remain the same regardless of whether private or public resources are in play. "The infrastructure needs to appear as seamless as possible to the applications," he says. Some of the applications aren't yet up to supporting such a mixed environment, he says.
Before the transition the company had single, rack-mount servers and is moving to blade servers sharing resources as a corporate pool.
He says he wants to make more use of Operations Manager for application performance monitoring that can identify where errors occur with more specificity, making it possible to resolve problems faster. Errors in customer facing applications could mean losing out on rentals, he says.
Read more about software in Network World's Software section.