One of the sneaky-big trends that gained momentum in during 2010 was the growth of converged infrastructure solutions. These are next-generation answers to more flexible data centers and cloud service platforms. Designed to leverage virtualization at all levels, such solutions bundle compute, networking, storage, and hypervisors into modular rack solutions. With so much in a single footprint, these solutions have earned the nicknames “Data Center in a box” or even “Cloud in a box.” Common examples include the VCE Coalition’s Vblock (a joint venture of EMC, Cisco, and VMware), IBM’s BladeCenter, and Dell’s PAN System.
Converged infrastructure is finding interest for data center refreshes but some of the greatest demand is among new and emerging cloud service providers, because it affords extreme flexibility in forms that are advantaged both from a capital and operational expense perspective. A growing number of managed services providers are launching Virtual Private Cloud services based specifically on these systems, and EMA expects this trend to continue.
A compelling opportunity presented by converged infrastructure is on the management tools front. All of the common management domains are present within a single footprint in these solutions, and they represent the first generation of data center architectures built from the ground up to optimize use of virtualization technologies. As such, they are stateless and dynamic, and thus present some real opportunities to achieve efficient and effective converged management.
The first challenge for managing converged infrastructure will be configuration and provisioning. Unified change and configuration management, coordinated by service construct, would be preferable here, although at worst it is certainly possible to continue to use element management tools for each domain component. There are a couple of integrated tools available – for instance the EMC Ionix Unified Infrastructure Manager covers all elements with VCE Vblocks, and Dell’s PAN Manager covers everything within that system. Alternatively, tools architectures that were put together to support cross-domain provisioning on non-converged infrastructures could also be used, if they happen to be in place.
The next opportunity, and perhaps the more compelling one, is to deploy unified monitoring of availability and performance. This is especially important due to the fact that all of the resources needed to deliver a cloud service will likely be resident within a single physical enclosure. Consequently, physical systems monitoring needs to be tightly aligned with virtual resource monitoring, again tied back to the applications or services being delivered. Again, it is possible to use standalone tools to monitor each domain type, but the real gains come when management systems can see the whole picture collectively, including the elements and the relationships between them. One current example of such a solution is the Vblock support recently announced by Nimsoft – a more detailed discussion of that solution is the subject of an EMA webinar scheduled for Thursday, December 16th – you can registerhere.