From Silos to Systems Management – Towards a Holistic Virtual Infrastructure
The argument is as old as the datacenter itself: should I buy infrastructure management tools from multiple, “best-of-breed” vendors or should I invest in a platform with integrated solutions? “Best-of-breed” products have the advantage of depth of functionality because their manufacturer only has to focus on one thing. However, tools that work seamlessly together because they share a common foundation have been shown to decrease operational costs and improve organizational effectiveness. The right approach, as you might expect, depends on the circumstances.
When it comes to managing the virtual infrastructure (and by extension the cloud) we are just beginning to weigh this decision. Until recently, most organizations have believed that the hypervisor vendors would provide all the necessary management tools for administering the virtual infrastructure. And to some degree they have -- especially for smaller, less “mission critical” situations. But as datacenters become more and more virtualized, they have reached the limits of tools supplied by the hypervisor vendors. As Denise Dubie, writing for Network World pointed out last May, “The time for a virtual systems management strategy has come. IT managers with growing virtual environments – servers and desktops – can no longer depend solely upon the management capabilities provided by their hypervisor vendor.”
Today’s tools from the hypervisor vendors struggle to handle larger, more complex deployments. These deployments typically require multiple instances of the hypervisor management console and often have no comprehensive way to merge and correlate information across virtual clusters.
Research from the Gartner Group suggests that IT organizations begin investigating virtualization management tools, beyond the hypervisor management console, when deployments grow larger than 100 virtual machines. At that scale, virtual segmentation, security, compliance, change control, configuration management, performance monitoring, capacity analysis, provisioning and automation become essential to the success of the virtual datacenter. While the hypervisor vendors claim solutions in all of these discrete areas, their tools usually only provide basic functionality and become isolated, “silo’ed” implementations. They lack the scalabilty, granularity and flexibility organizations have become accustomed to in their physical datacenter.
This is why many IT organizations have looked to Enterprise Systems Management (ESM) providers for help. They have been providing “industrial strength” systems management tools for many years. However, so far the major ESM vendors have few purpose built tools for the virtual infrastructure. Existing ESM tools lack the understanding of hypervisor and hypervisor console configuration. They find it difficult to keep up with the highly dynamic nature of the virtual infrastructure - hundreds of configuration changes per minute are not atypical in even a modest virtual datacenter. ESM vendors may adequately track “internal” server configuration parameters (Windows, Linux, etc.), but don’t understand “external” server parameters such as CPU, memory, disk size, virtual network interfaces (vNICs), and storage mappings since these setting have, previous to virtualization, been physically configured. Tools from different ESM vendors may be an effective way to manage servers, networks and storage but they tend to break down in a virtual infrastructure because the key elements are all virtual objects which exist only in the hypervisor.
