Posted by: Beth Cohen
Cloud architectures, enterprise architectures, enterprise cloud services, enterprise IT, OpenStack, VCE, VMware
Question: Building a private cloud can be an overwhelming project involving a holistic understanding of operations and development. Is it really possible to shortcut the process?
Not surprisingly according to just about every research study out there, including the Open Data Center Alliance recent survey of its members, enterprise cloud is the next BIG THING. This survey predicts an adoption rate five times faster than predictions from other research firms such as IDC and Gartner. Of course, the numbers could be skewed by the self selected nature of the participants. I do agree that the enterprise needs to take it seriously and move their IT portfolio to a cloud model. Every indication is that most enterprises are mapping their cloud strategies and seriously investigating how to do just that. However, anyone who has tried to build an enterprise cloud knows how hard it actually is. So for enterprises anxious to move to a private cloud that do not want to take the fifteen years and thousands of engineers that it took Amazon, there are a plethora of companies that are jumping on the enterprise cloud bandwagon with simplified “cloud in a box” type offerings. Can these premium priced products deliver on their promise to get an enterprise on a cloud platform that much faster or are they just marketing rebranding?
Most of the cloud-in-a-box products are from the usual big IT services shops. Many of them, such as BizCloud from CSC and VirtualSystem from HP are based on VMWare with some combination of standardized network, storage and systems hardware. vFabric from VMware seems to be more a kitchen sink of VMware owned “cloudy” products than a vertically integrated service such as Azure. Cloudburst from IBM and VCE of course could both be considered cloud-in-a-box products, but they are hardware only integration and both come with big price tags that in my mind defeat the purpose of why companies move to the cloud to begin with — which is to save money.
That VMware is the underlying hypervisor of choice is not overly surprising since VMware owns 90% plus of the enterprise data center virtualization market already. The enterprise is comfortable with their reliable and mature virtualization tools, so that approach would seem to be a natural fit for risk adverse organizations that want to move to the cloud as painlessly as possible. For enterprises who do not want to pay the VMware or the big vendor premiums, there are some interesting tools from StackOPS, Dell (Crowbar) and RackSpace based on the Open Source OpenStack project. So far these tools are more recipes and automation utilities designed for enterprises that have access to good dev/ops resources and are willing to live on the enterprise cloud leading edge.
The fallacy is that cloud architectures are not just virtualization with some added bells and whistles. It is a different way of thinking that requires enormous changes to the enterprise IT organization, because it requires a different understanding of how IT services are delivered at scale. The only way to realistically support tens or hundreds of thousands of systems is to automate everything; and the only way to achieve the level of automation required for a successful enterprise cloud deployment is for development, test, integration and operations to become part of a continuous process, not the separate functions they are in traditional organizations. In the end no matter how big a check you are willing write, the current crop of cloud-in-a-box software and hardware offerings are not addressing the fundamental organizational differences between cloud infrastructure operations and IT infrastructure business as usual.
About the Author
Beth Cohen, Cloud Technology Partners, Inc. Moving companies’ IT services into the cloud the right way, the first time!