Server virtualization gives organizations an opportunity to make their data centers energy efficient by reducing their physical server footprint -- sometimes by a factor of 20 to 1 or more. After implementing server virtualization, enterprises can expand the technology to other layers within the data center, including desktops and applications. However, once you've implemented server virtualization, you become dependent on a single solution, as each host server runs multiple production workloads. Organizations can't afford to lose that server to failure; it has to run. If it doesn't, though, it must be able to hand off its workload to another standby host so that production applications remain available at all times.
It's important to look at the different ways to protect production workloads and ensure that they're always available. Here are three different methods for making workloads contained within virtual machines (VMs) highly available:
1. Create host clusters.
2. Create guest failover clusters.
3. Create a guest server cluster or network load balancing (NLB) clusters.
Each of these options can provide a specific level of availability to VMs and to the applications they run. By using high-availability (HA) cluster configurations, here are some potential issues you can address in your environment:
- Protection from host failures. When a host fails, either the VMs that it runs will be moved off to other hosts or applications within the VMs will be moved to other VMs. Which scenario occurs depends on which of the three HA strategies you use to protect a host.
- Host server maintenance. Production applications will continue to run because other VMs will support them or because the VMs will be moved.
- Workload migration. In the data center, you can dynamically move application workloads from one machine to another. This ensures that applications run at optimal performance -- even during peak periods.
Each of the three methods – host clusters, guest failover clusters and guest server clusters or NLBs – is available when running a hypervisor; however, the method that's required for each application depends on its level of importance. In most organizations, nearly all users need financial systems or email, but few users will miss a network maintenance server that fails for a short period of time. Consider each application type to identify the best HA method for your enterprise.
Danielle Ruest and Nelson Ruest are IT experts focused on continuous service availability and infrastructure optimization. They are authors of multiple books, including Virtualization: A Beginner's Guide and Windows Server 2008, The Complete Reference for McGraw-Hill Osborne. Contact them at firstname.lastname@example.org.