Guest contributor Lynn LeBlanc considers the challenge of managing multi-hypervisor environments. Is there an approach that minimizes the complexity?
By Lynn LeBlanc, CEO of HotLink Corporation
Virtualization dangled enticing promises in front of IT, and many shops grabbed them, moving more than half of all server workloads to virtual assets as of 2012, according to industry experts. Server virtualization is obviously on the rise, and it has delivered on its promises - dynamic resource management, improved business continuity, and the cost savings that come from server consolidation - but not without some challenges. VMware was there at the beginning of the rapid transition from physical to virtual environments, and its vSphere is generally the dominant hypervisor in most large shops today. However, as virtualization spreads to encompass the next wave of applications - beyond the business and performance-critical ones -more organizations are embracing multiple hypervisors to keep costs under control. So what does the multi-hypervisor data center look like?
Heterogeneous hypervisors and the march toward tiering
It's broadly acknowledged that standardizing on VMware is cost prohibitive for fast growing enterprises. As data center managers work to virtualize more workloads, they are shifting less-critical assets to lower cost virtual infrastructure like Microsoft Hyper-V, Citrix XenServer, and Red Hat KVM. This reserves pricier VMware vSphere licenses for mission-critical workloads. In many ways, this mirrors exactly what happened with storage arrays. Organizations used to buy monolithic storage arrays to meet general purpose needs. Today, these same organizations optimize budgets by implementing multiple tiers of storage from different vendors and manage them through the use of integrated tools. This tiered approach to optimize price/performance works for server virtualization -- but requires a unified management framework to realize the benefits.
Traditional approaches to multi-hypervisor management (and why they fail)
All of the benefits of pervasive virtualization and tiered virtual infrastructure come with some read-the-fine-print realities. While hypervisors are functionally very similar, providers have integrally coupled advanced and multi-host management capabilities with specific hypervisors. As a result, virtualization platforms each have native consoles for administration and management, and managing multi-hypervisor deployments can create a web of management tools that simply do not work together. Of course, hypervisor vendors provide APIs for integration and overarching management, but they only expose remedial features. There is a long list of tools available that use these APIs for various overlay automation functions like service catalogs, orchestration, and self-service provisioning, and each highlights support for multi-hypervisor platforms. However, these overlay solutions still require deployment of the native consoles for administration of each virtual platform supported. Basically, this is a manager-of-managers model with a high level of implementation complexity, cost, and questionable value for many.
As a result, many enterprises default to operating the various virtual infrastructure platforms in silos. There's no benefit of increased agility or dynamic resource allocation, but at least it's manageable on a day-to-day level.
A better approach to multi-hypervisor diversity: Transform IT
Without a simple, scalable and seamless way to manage multi-platform deployment, system administrators will either buckle under the pressure or simply walk out the door in search of greener pastures. To efficiently manage a multi-hypervisor environment, IT requires a solution that addresses administration, deployment and day-to-day operations in a practical and truly unified model -- ideally, one that is an extension of the management skills and infrastructure already made. Instead of providing specialized automation on top of multiple native management toolsets, IT managers should focus on options that allow the abstraction of hypervisors and workloads within their existing enterprise virtualization management infrastructure -- namely VMware vCenter.
With the latest transformation technologies, VMware vCenter can be extended to be the single point of administration and management for multi-hypervisor environments and even a single point of integration for all virtual platforms -- both on and off-premise. With transformation, administrators accomplish the same task on two or more platforms in exactly the same manner using existing management methods. No other consoles, toolsets, APIs or skills are required. Cross-platform cloning, snapshots, templates, migrations and workload conversions are all enabled. Non-VMware resources, like Hyper-V, XenServer, KVM and even Amazon EC2, are managed seamlessly and uniformly alongside vSphere, utilizing the underlying capabilities of VMware vCenter.
With the uniquely simple, bottom up transformation approach to enabling the administration, management and integration of heterogeneous environments using existing management technologies and processes multi-hypervisor management in the data center emerges a clear winner in 2013.
Lynn LeBlanc, CEO and founder of HotLink Corporation, has more than 25 years of enterprise software and technology experience at both Fortune 500 companies and Silicon Valley start-ups. Prior to founding HotLink, LeBlanc was founder and CEO of FastScale Technology, an enterprise software company acquired by VMware, Inc.