Executive Viewpoint 2017 Prediction: Cirba – The Year of Cloud-Washed Virtualization
With public cloud gaining momentum 2017 will be the year that many organizations will get down to the dirty business of figuring out how they can actually migrate their precious enterprise apps into the great data center in the sky. Development groups are no stranger to public clouds, where PaaS technologies and micro-service architectures have helped them to develop a new breed of cloud-native apps. But this approach doesn’t help with the mountain of legacy apps that most organizations are saddled with, and their more monolithic design.
In the latter half of 2016 VMware and AWS announced a new offering will likely be very useful for these applications. The upcoming ability to rent VMware ESX hosts from Amazon, and not just virtual instances, will create a very interesting dynamic when it comes to cloud migration and hosting decisions. It essentially puts the power to overcommit back in the hands of the consumer, and allows workloads to have complementary utilization patterns to be combined on physical hosts in a way that optimizes the sharing of resources. This “workload stacking” capability was arguably one of the biggest drivers of virtualization in the last decade, and enabled density gains that fueled tremendous cost savings. But leveraging this capability in the cloud changes the nature of cloud management, and customers won’t just have to worry about sizing their cloud instances to control costs (which has been a major issue for most), but will also require a means of ensuring they are operating these new environments efficiently.
The downside of this bare-metal strategy, however, is that it arguably isn’t really cloud at all. It is more of a co-location strategy masquerading as a cloud, and aside from a shorter-term rental model it provides none of the rich services and marketplace features of the standard cloud offerings. But this may not matter, as it is really more of a branding exercise than a technical innovation, and if the service provider is branded as a “cloud” then running workloads there must mean you “moved them to the cloud”, regardless of exactly how they are hosted there. And when executives pound their fists demanding that cloud be adopted, infrastructure groups can give a big thumbs up. Clever indeed.
In 2017 this will likely prove to be a very useful strategy, and will form an important part of the overall portfolio of hosting assets, along with traditional public IaaS and next-gen on-prem virtual environments. The challenge facing most organizations will be which apps and workloads should be hosted using each strategy, and how to optimize the density across all three to minimize total cost while at the same time getting the benefits of reduced management complexity and increased agility. Unfortunately, the vendors providing these hosting technologies do not provide this type of optimization, be it workload-driven migration analysis, multi-cloud hosting venue optimization, or workload densification. And any claims that they do should be met with skepticism, since it would directly impact their business. But for organizations that can get a grip on this by understanding their true app requirements and workload patterns, and use this to scientifically determine the best way to host them, 2017 will provide even more opportunity for successful cloud migration.