David M. Smith of Gartner spent some time with me a few weeks ago exploring Cisco’s cloud computing vision. I walked him through a variety of concepts, such as private clouds, virtual private clouds and open clouds. We also spoke of the vision of an “Inter-Cloud”, a vision shared by most cloud enthusiasts here at Cisco, including our CTO, Padmasree Warrior. David had some excellent feedback, and I enjoyed the conversation very much.
Unfortunately, I think David came away from the conversation with a slight mis-interpretation of how we use the term Inter-Cloud. As he states in a post on his blog, “Life on the Inter-cloud”:
A few weeks ago, Cisco’s James Urquart (sic) told me about their cloud strategy and use of a term ‘the inter-cloud” as the evolution of private clouds connected together. My mind flashed back to 1996. While doing my daily skim of the interblogs today, I came across A Hitchhiker’s Guide to the Inter-cloud and the flashback recurred.
I do not mean to diss James or Cisco — I believe they are indeed, well meaning, and indicative of a view of cloud computing that evolves from the infrastructure/data center perspective. And it is yet another example of why terminology matters and why we talk about there being one cloud ( a euphemism for an abstraction ) and frown upon use of “clouds”, in the plural sense. The world of cloud computing has become so cloudy and the terms are becoming so confusing that I have to wonder if we will ever get past it.
I posted a response in the comments to David’s post that I think clearly lays out the distinction between private clouds and the Inter-Cloud vision. But I’m too close to the topic to be completely objective. I copied my response below; please read it and let me know what you think. Did I help clarify the terminology we use at Cisco (and that is being adopted elsewhere) or did I just muddy the waters further? Any feedback is welcome.
I think you misunderstand our use of the term “Inter-Cloud”; we use it in a way that is completely parallel to the term “internet”, not “intranet”. Let me break it down for you:
Today, people are creating individual online services that can be accessed “at scale” over a network on an on-demand fashion. These are “clouds”
The world of public clouds is growing rapidly to include many offerings across the SPI model–SaaS, PaaS, and IaaS.
Enterprises are building (and will continue to build) “private clouds”–clouds in which they maintain control of resource usage and trust boundaries across both internal and external resources from a single management system. Initially private clouds will be made up of almost entirely internal resources. “Private clouds” are a good analogy to “Intranets”, to be sure, and there will be many of them.
At some point in the not horribly distant future, some service providers will offer “virtual private cloud” services to allow “private clouds” to consume resources in the service provider infrastructure, while maintaining the illusion of being a part of the customer’s private cloud. This is simply extending “intranets” to consume services over the Internet without exposing the content to the general public–kind of like VPN? (Not a perfect analogy, to be sure.)
In the meantime the set of public cloud services evolves, standardizes, and becomes a more open market. Not all will be virtual private clouds services; there will be other forms of interoperability. These sets of interoperable, interchangable clouds could be thought of as “open clouds”.
In the early stages, however, there will be relatively tight coupling between the enterprise and any individual public cloud offering chosen; not necessarily lock-in, but the time taken to make a change is still somewhat onerous and involves direct agreements between the customer and the vendors involved. So “open clouds” are not yet the most elastic markets they could be.
The network technology to enable the linkage of enterprises to all forms of public cloud offerings (not just virtual private clouds) in a way that takes the unique nature of cloud computing and running IT workloads in mind is called “cloud internetworking”.
The final phase–many years from now–includes the introduction of publicly shared core services–very much like DNS and peering–into the carrier networks that enable a more loosely coupled relationship between customer and cloud vendors. This serves to greatly increase the elasticity of the cloud market, and creates a single public open cloud internetwork–the Inter-Cloud.
To be sure, we are still in the very early stages of envisioning both the nature of the Inter-Cloud, and the technologies required for its formation. We do not, however, confuse “private clouds” with the Inter-Cloud.