The Understanding Problem
While presenting a keynote address to an audience of information-technology people from the higher-education sector, a senior representative of a super-vendor smiled handsomely and said:
“The cloud means different things to different people.”
That is a very unhelpful statement. Perfectly-clear criteria define what cloud computing is, and therefore what it is not. There is still widespread marketplace confusion about the benefits, patterns, and use-cases of cloud computing. This confusion is caused partly by people lacking or losing reference to the fundamental definition of what cloud computing is.
The Elastic Definition
In 2010, the Burton Group defined cloud computing as:
“The set of disciplines, technologies, and business models used to deliver IT capabilities (software, platforms, hardware) as an on-demand, scalable, elastic service.”
This definition is as good as any that i’ve seen, and it is accompanied by a set of five essential characteristics, features of a service that must be present if the service is able to be defined as cloud computing:
- it uses shared infrastructure
- it provides on-demand self-service
- it is elastic and scalable.
- it is priced by consumption.
- it is dynamic and virtualised.
The NIST definition published in 2011 also includes essentially the same set of five essential characteristics that must be present for a service to be described as cloud computing:
- on-demand self-service
- broad network access
- resource pooling
- rapid elasticity
- measured service
Both NIST and Burton go on to describe service models (e.g., IaaS, PaaS, SaaS) and deployment models (e.g., private cloud, hybrid cloud, community cloud), but the defition and the essential characteristics must be present for a service to be a cloud computing service.
Abolishing the Confusion
Abolishing the confusion should be easy: engage with a clear sense of the essential characteristics that must be present for a service to be considered as a cloud service… and be equally clear about when it really matters whether or not a service falls into the official definition of cloud computing.
The CloudU initiative curated by Ben Kepes (@benkepes on Twitter) with Rackspace is a good resource for raising globally the level of cloud understanding. CloudU is part of the Rackspace knowledge centre at http://www.rackspace.com/knowledge_center/cloudu/
Some other selected useful references are detailed below:
- Blakley, B. & Reeves, D. (2010) Defining Cloud Computing, Burton Group Research
- Kaskade, J. (2009) Defintion of Cloud Computing — Again, Jim Kaskade’s blog, http://jameskaskade.com/
- Mell, P. & Grance, T. (2011) The NIST Definition of Cloud Computing, National Institute of Standards and Technology, U.S. Department of Commerce, NIST Special Publication 800-145, http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdf
The emergence and the importance of the role cloud-service brokerages have must be achieved against a complex backdrop that is difficult to negotiate, as it includes:
- cloud roaming
- data sovereignty
- throughput expectations
- reliability requirements
- scalability decisions
- geographical considerations
- costing models
- security capabilities
- interoperability needs
- integration needs
- bandwidth availability
- time-of-day for follow-the-moon provisioning
…and the necessary standards do not yet exist for cloud services to be truly portable, leading to potential issues with cloud-vendor lock-in, cloud migration, and cloud federation.
When they do arrive and are formed fully, what will cloud standards look like? One visiting cloud analyst suggesting the following:
- cloud standards will be reasonably low-level
- thinking about restaurants, the low-level standards and equipment for a McDonalds are very similar to those of The French Laundry
- the experience of eating at a McDonalds is very different from the experience of eating at The French Laundry
Cloud standards will exist more at the carrier/provider level than at the user/customer level, and considering the full raft of quality-of-service needs will be essential. The extent to which the NIST and the CSA and the analyst firms and the cloud providers achieve usable cloud standards, and when they deliver them, is yet to be seen.
There’s a lot to like about the CloudU initiative curated by Ben Kepes and sponsored by Rackspace.
Increasingly, use of the C-word generates cringes because there is a still-developing perfect storm of confusion, hype, and fear around The Cloud:
- cloud solves all our problems
- cloud is a security nightmare
- cloud is anything off-premise
- cloud gives all your stuff to the government
- cloud is mature and ready for anything
- cloud is anything virtualised
- cloud is anything in the bucket of things i don’t understand
- cloud integration is easy, reliable, and standards-based
…and the CloudU initiative goes a long way towards establishing a good basic scratch-the-surface level of common understanding. There’s a good case here for every person who uses the word “cloud” needing mandatory CloudU certification.
Some of the quiz items are either generic-technology (e.g., “What is the purpose of a load balancer?”) or have a standout bad answer (e.g., “Which is not a popular type of cloud service pattern: public cloud, private cloud, hybrid cloud, or big-fluffy-cloud-in-the-sky) and there’s a some repetition across the modules around topics like API, but they do serve their purpose.
Beyond the quiz items, the reference/study guides and the textbooks are well-written and cover sensibly-chosen topic areas, providing a balanced view of the pros and cons of choosing cloud deployments.
Those wanting to take the next step might work through the NIST Cloud Computing Defition before heading into the industry analysts (e.g., Diversity, ZapThink, Gartner) and the vendor whitepapers and the academic-leaning treatments of cloud computing.
Meanwhile, i enjoyed the CloudU experience greatly, and am pleased to display my certificate badge here!