Find a way forward in the cloud vs. virtualization debate
IT admins weighing the options available for cloud and virtualization services against use cases and economics must navigate a complex market with an uncertain future.
The cloud vs. virtualization debate -- the choice between running workloads in the cloud or on traditional virtualized servers -- isn't simple. IT admins must consider all the points of a three-sided debate about public clouds, private clouds and virtualized clusters. Each approach is at a different stage of maturity and offers different value propositions.
Public cloud code cores are mature, and cloud service providers continue to release services at a prodigious rate. This is a strong case to use the public cloud for tenants who are comfortable running workloads in a multi-cloud environment.
Even so, some organizations might want to keep parts of the total service in-house in a hybrid cloud model. Unfortunately, this is the least mature segment in this market. The industry lacks seamless public-private management and data management is poor and slow. Worse, the only private cloud stack that is currently usable is OpenStack, which has many followers, but remains relatively immature.
Companies such as Rubrik address the data management challenge, whereas the major cloud service providers compete for seamless management, albeit in very different ways.
Major vendor offerings
With Azure Stack, Microsoft plans to make Azure both a private and public stack, which should result in complete cloud portability. It could take several months after Azure Stack's release for Microsoft to clear up performance and cloud bursting automation problems, but this is a powerful direction for the cloud and virtualization markets. Windows colors Microsoft's world, however; Linux and Hadoop shops might struggle to integrate Azure Stack because of that.
The industry lacks seamless public-private management and data management is poor and slow.
Google is working with OpenStack to unify scripting and image libraries. The result should also be homogenous, except for the initial cloud setup, which will be simplified over time. Google's focus is on offering the fastest cloud for Linux and big data apps.
AWS is building huge private clouds for governments and offering dedicated cloud spaces as alternatives to in-house clouds. However, this isn't enough for many customers that want an in-house, controllable facility.
Rather than compete with VMware to bring the AWS cloud in-house, VMware and AWS partnered up to enable vSphere control in AWS public cloud instances. This makes the cloud an extension of in-house virtual clusters, with the benefit of on-demand scaling without the massive rewriting of scripts and retraining of staff.
Admins face a divergent path in the cloud vs. virtualization debate
So how should administrators choose between cloud vs. virtualization? They must consider the infrastructure their organization already runs, the price and their expertise.
Figure A. Choose between cloud and virtualization features.
Admins who are interested in extending an existing VMware facility to support cloud bursting can keep their VMware technology and follow AWS for cloud operations. The upsides to this strategy are business continuity and protection of the VMware investment. The downside is that cloud-like automation will become more necessary with time, and this strategy must address that. Integration of VMware and AWS' long list of services is also necessary.
Power users will probably want to go the Google and OpenStack route, which is optimal for performance and analytics. With this option, the alternatives of using a virtualized cluster or bare-metal hardware remain open, but Google will increasingly offer analytics services for rent to make in-house propositions less tenable.
Economics are also a significant factor. The price of public cloud instances continues to drop, which adds a new dimension to cost comparisons with a fixed cost, upfront purchase of in-house hardware. It's easy to underestimate the number of instances necessary for a particular project, but a cloud sandbox can make the cloud vs. virtualization calculation easier by helping admins estimate instance efficiency before making a full commitment.
Still, a transition from in-house computing -- virtualized or not -- to the cloud is costly in terms of training and business disruption. This might be enough to convince some admins to stay with virtualized servers after weighing the cloud vs. virtualization options, but VMware admins might see AWS as an option for a soft landing.
