IT practitioners make gut decisions that effect the performance and reliability of complex assemblies of hardware, software, and applications. The survivors usually call it right.
The business case for VMware and other hypervisors for small applications running on lower performance hardware is overwhelming. Utilization of server, storage, and network components is improved and less hardware is required. As a bonus, there is much greater flexibility. Added functionality such as a good management tool suite that can reduce IT management costs is the icing on the cake, but unless the additional expenditure is easily justified, the workloads in general are not going to significant additional implementation costs and complexity. Minimal management is all that will be justified.
The business case for virtualizing a large-scale mission-critical application supporting thousands of users with a complex, latency-sensitive database is not so obvious. The utilization cost savings are still there, and hypervisor management suites can make the business case for virtualization significantly better. But the incremental saving on infrastructure and staff costs will not convince a seasoned server, storage, or database specialist to virtualize that application. The specialist is likely to say “the application will run faster on bare metal, I can diagnose problems more easily on bare metal, I can provide higher availability by clustering bare metal, and my business continuance environment will run better on bare metal.” He will argue that the risks to the business of system failure are much higher than the marginal hardware, software, and staff savings justify. And it is a brave business or IT senior manager who will call him. The result is virtualization stall, and a few systems that may be consuming as much as 50% of the infrastructure budget.
The key point is that Hypervisor Management suites are comparing themselves against the wrong environment, tilting at the wrong windmill. The question is not how much better the management suite makes the virtualized environment, but how much better it can make the virtualized environment compared to a bare metal environment.
Elements of such a business case should include:
- The ability to put the latest and greatest resources in support of the most critical components of the critical applications hardware dynamically and automatically. This can allow more resources to be applied critical applications in response to business change or infrastructure malfunction than dedicated hardware.
- The ability to enhance availability, recoverability, and security over dedicated bare metal systems,
- The ability to enhance the applications and infrastructure more quickly and with less risk than dedicated systems.
- The ability to report total application performance and resources consumed to the business more accurately that a traditional SAN environment with dedicated servers.
Growing an Infrastructure 2.0 hypervisor environment out of the stall depends on providing solutions that will answer these challenges to the satisfaction of senior IT professionals.
Action Item: Ask hypervisor management vendors (Citrix, IBM, Microsoft, and VMware) and management software providers (including the vendors such as CA, HP, IBM, Microsoft, Vkernel, VMTurbo, and others) for a roadmap and timescale to provide better performance, security, and availability for a clustered hypervisor environment than dedicated hardware.