Published: 02 Jul 2014
Converged infrastructures assemble components from multiple IT domains into one integrated system by combining virtualization, storage, networking and compute. A management framework is also included that allows orchestration of the entire infrastructure so that each constituent technology doesn't have to be managed separately. But from that set of basic parameters, the individual flavors vary significantly, and those different flavors are well worth a closer look.
Throughout 2012-2013, we watched converged infrastructure with some degree of skepticism, with doubts about whether it would really take hold. After all, we've seen many different flavors of converged IT functionality come to market with many failing to gain much attention. But as we entered 2014, the notable successes of a few vendors left no doubt that the convergence trend is healthy and here to stay. The VMware-Cisco-EMC coalition – VCE -- topped $1 billion in revenues, and rapidly growing Nutanix appeared to crest a $100 million revenue run-rate somewhere around mid-2013. Meanwhile, every major storage vendor is offering some form of converged infrastructure and the startup ranks are well represented with other innovators such as Scale Computing and SimpliVity.
Why is converged infrastructure taking the market by storm? This idea of converged infrastructure is rooted at the intersection of a number of different trends. It used to be that, given the available technology, IT was forced onto a path of sprawl and inefficiency. We routinely reacted by buying more gear and adding to the sprawl. But in the past three years, a bevy of new technologies have become available that should allow us to reverse this trend:
- Solid-state has all but solved the storage performance bottleneck, making it possible to squeeze out performance efficiencies from storage systems.
- Virtualization has simplified the networks and pipes in our data centers.
- The cloud allows us to store enterprise data with better and more ubiquitous access, while pushing much of the low-priority data that contributed to sprawl into a flatter, more scalable and often outsourced infrastructure.
- WAN optimization, deduplication, data transmission, dispersal and caching enhancements have all helped these other technologies work a bit better.
Meanwhile, all of these innovations have come together just when IT had seemingly reached its limit, with many shops running up against hard limits on inflexible things such as available data center power or capital dollars.
Traditional methods are failing
The reality, though, is that the old ways of doing things had already crossed the line of practicality. We've been suffering through escalating complexity for a number of years, so these new technologies have created a significant shift in IT expectations. With the new choices and their ability to solve serious data center problems, technology practitioners have decided that it's time to forgo temporary fixes and patches. We need to apply new technologies to lifecycle problems, from deployment to decommissioning. What was once a never-ending exercise of trying to fit pieces of disparate technologies together, is giving way to the host of integrated technology offerings now available.
Converged infrastructure delivers on promises, such as providing a right-sized, completely packaged solution that could be ordered as one item. When that pre-integrated and largely pre-configured item arrives in the data center, it can effectively put an end to complex deployments and configuration exercises. And once deployed, that converged infrastructure operates as a unit, giving users a single point of management for all the underlying components, as well as application-layer elements that may also be packaged in the solution. But perhaps more importantly, converged infrastructure solutions promise easy repeatability as well. When additional infrastructure is needed, users can simply order another unit. In building block fashion, that additional unit could then be deployed alongside the existing system, with both managed under the same umbrella.
Convergence spawns innovation
Convergence is also developing into a hotbed of innovation. One of the shortcomings of this first generation of converged infrastructure products was that units didn't elegantly add to the whole without introducing new boundaries. These converged infrastructures still build on limited storage pools and complex network infrastructures that force users into more complicated workload partitioning and storage movement, with significant management complexity across components. All that added up to costs in the form of time and effort, troubleshooting and potential outages.
But a handful of vendors -- notably Nutanix, Scale Computing and SimpliVity -- have introduced offerings that are now referred to as hyper-convergence. Hyper-convergence does away with the partitioning and complexity of first-generation converged infrastructure. In effect, these vendors have reduced the building block to the smallest possible entity, and that block contains storage, networking and compute resources that add to the whole in a scale-out fashion. That makes it possible to start small and scale to a very large infrastructure with no apparent change in complexity or management.
Some of those vendors are growing rapidly. These disruptive changes are coming your way, and they're well worth paying attention to. Converged infrastructure and hyper-convergence are disruptive to traditional infrastructure vendors because these new products offer customers alternatives to improve data center density and efficiency. Some of the old guard is scrambling to adjust to the converged movement and, as a result, you're likely to find that the familiar components you're used to buying are fading away faster than expected. That's not a bad thing if you've planned for it. Have you paid attention to converged infrastructure and planned accordingly?
About the author:
Jeff Boles is a senior analyst at Taneja Group.