animind - Fotolia
A multi-hypervisor setup can help mitigate costs over time, but the complexity of implementing a multi-hypervisor environment can seem like a big hurdle to overcome.
For many admins who have exclusively relied on VMware's products, moving workloads to Hyper-V might be daunting. But, by splitting workloads up, admins can realize significant cost savings and continue to grow their environments.
VMware's ESXi hypervisor is seen as a mature and feature-rich product, but it comes at a high price point. Hyper-V's feature set -- although not as mature and extensive as VMware's -- has come a long way, and it's less expensive. This presents two hypervisor levels: one more suited for high-end workloads and one equipped for middle-tier workloads.
Mix test/dev and production
Admins might opt to put core production in VMware and test/dev on a Hyper-V host because it's a straightforward approach; it balances host licensing costs and effectively splits workloads. But, by doing this, admins create a situation where an outage can have a devastating effect on the data center. If all production workloads are in one place and they become compromised, the whole operation can go down.
By mixing test/dev with production -- at a 60/40 mix, for example -- the outage's affect is significantly reduced because admins have fewer production servers going offline. Mixing test/dev with production can seem unnecessarily complicated when trying to decide which workloads to put on Hyper-V hosts and which to keep in a VMware environment, but careful planning will make for a stronger infrastructure.
Consider priority and usage
Before admins implement a multi-hypervisor environment, they must take the time to understand their workloads. Production servers and test/dev are often categorized with little investigation into what workloads actually make up those categories. A larger modern application is made up of dozens of servers that perform a wide range of functions, and some of those functions aren't as critical as others.
Reporting and monitoring servers are application utility servers that are important but don't typically affect client-facing experiences, so they make good candidates for a second-tier production environment. Citrix and other virtualized presentation servers are often considered key for production, but admins should take a look at their typical usage before automatically placing them in the VMware environment to determine ideal workload placement.
For example, if an admin averages 80% usage with only a few spikes above that, moving the other 20% to a Hyper-V cluster won't be an issue because both hypervisors are production-capable; the production environment can handle those workloads with some considerations. If admins take this approach, they can place production support servers, additional Active Directory controllers, backup DNS and Dynamic Host Configuration Protocol infrastructure servers in the second-tier environment to help balance out costs.
Manage multi-hypervisor environments
Once admins have workload placement under control, they should look at management. With VMware, vCenter is the primary management platform because it enables admins to centrally manage both hosts and VMs.
Microsoft offers a few management options. Windows Server comes with Hyper-V Manager, and for stand-alone hosts, it works well. Admins also have access to Windows Admin Center, which builds on those features. However, if admins want to treat Hyper-V as a true production environment, then they should use System Center to manage it.
Admins need System Center in this case because vCenter lacks much of the necessary features. System Center is a suite that contains several management products, such as System Center Configuration Manager and System Center Operations Manager. If admins want to use vCenter instead of System Center, they'd need to purchase other products, such as vRealize Operations, to realize the same range of capabilities.
Every environment is different, and as such, management needs differ. But, in general, VMware's management software is a la carte, while Microsoft offers more of a total package.
A single management platform for multi-hypervisor environments is still a ways off, so admins must be flexible and open to multiple management interfaces. Some additional training might be required, but the benefits are worth it.