While cloud computing can be many things to many people, two things about cloud almost universally apply: It's a catalyst not only for IT efficiencies, but in a larger sense cloud is also forcing changes surrounding the very role of IT. This is because IT executives are being asked to account for value and broker services more quickly and responsively than in the past, or else risk being bypassed altogether as lines of business seek cheaper or more readily available services.
In this environment IT executives need to be able to understand, define and support critical business priorities, while managing services that will often have a mix of internal and external interdependencies. This requires establishing baselines for monitoring and managing a tangled web of vendor relationships and internal initiatives without getting lost in a sea of warring, and sometimes competitive, organizations, politics and technologies.
For related articles, visit Internet.com's new cloud computing site.
This series will include some tips for establishing baselines, identifying roadblocks, and setting measurable objectives as IT organizations move towards cloud. One hint -- in spite of the marketing language of more than one infrastructure-related vendor cloud is not the endpoint of "a journey," but rather a mix of enabling technologies that, if used wisely, can support more flexible, responsive, cost efficient and resilient service delivery.
12 questions to ask for optimizing cloud deployments
One of the most effective ways to get good answers is, of course, to ask the right questions. Good questions followed up by meaningful actions are often more conducive to success in IT than trying to implement a pre-established catechism of "best practices" that typically allow IT organizations to avoid true self-examination for boilerplate solutions that work well only as boilerplates.
So here are twelve questions to help you set your organization in motion around a strategy for cloud.
- Why are you doing this? What is cloud computing and why do you even want to bother with it?
- Where should you begin?
- What are you doing now? How can you establish a meaningful baseline?
- Should you look for an internal cloud strategy or external cloud services?
- What phase one objectives for improvements do you think make sense given this baseline?
- Do you need to make organizational or process changes to support cloud?
- Do you need to invest in new management technologies to minimize risks and optimize cloud?
- If you are looking at all externally, what should you look for in a cloud service provider?
- What SLAs apply in support of your Phase One objectives and how does cloud make things different? How can you realistically set meaningful SLAs, for instance, across a combination of internal and external interdependencies?
- How do you measure your SLAs, which are usually performance oriented, against costs? If cloud is on-demand, how do you account for usage and value?
- What about the business impacts of cloud beyond IT? How can gain visibility-- and measure -- if cloud adoption helping or hurting the broader business or organization you support?
- How do you know if Phase One is complete and/or successful? What should you do next? What are your next step objectives?
Getting started: why are you even doing this?
Just in case you havent already noticed, one of the challenges surrounding cloud is the fact that the term cloud is, well, cloudy. Im certainly not the first to say this and the confusion surrounding the ironically appropriate term has produced countless white papers and webinars that in the end have a kind of medieval/academic flavor (e.g. when they used to debate things like the eight spheres that move around the earth, and the duplicity of the moon).
So Ill try to keep this part rather brief and to the pragmatic point at hand which is, to put it bluntly, why should you even be bothered?
The U.S. National Institute for Standards and Technology (NIST) defines cloud computing as: a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.
Not wanting to add to the general debate, EMA has pretty much stuck to this one. Although Ill put in my two cents and say that my one reservation about NIST is that once you read further it stresses a bit too much what the consumer of cloud services cant do -- reinforcing an opaqueness surrounding cloud that is probably unhelpful and certainly not what most in IT want at least given our research.
A few salient features implied from this definition are cloud computing is:
- On-demand self service: so that users can in theory unilaterally provision computing capabilities or access application services when needed.
- Broad network access: makes cloud services available to a larger number of users across different device types (e.g. mobile phones, laptops, PDAs, etc.).
- Resource pooling: typically requires dynamically assigning different physical and virtual resources based on customer demand. This could mean dynamically reallocating systems, storage, network bandwidth and applications based on user needs in a way that makes the physical and geographical location of the resources invisible to the user.
- Rapid elasticity: so that resources can be quickly and easily provisioned based on demand.
- Measured service: so that cloud services can be accurately accounted for and assessed in terms of cost and value with transparency to both provider and consumer.
NIST also defines three categories of cloud services that can apply internally or externally:
- Software-as-a-Service (SaaS): Providers deliver applications to consumers over a cloud infrastructure typically accessed through a thin client interface, such as a web browser.
- Platform-as-a-Service (PaaS): Here the provider offers both an underlying cloud infrastructure and application-related tools and resources such as programming languages to an internal or external consumer.
- Infrastructure-as-a-Service (IaaS): Providers deliver cloud infrastructure support (systems/processing, storage, networks, etc.) to support consumer (internal/external) requirements for running applications and operating systems.
SaaS is the dominant perceived use of cloud, although many SaaS deployments do not live up to all of NISTs requirements -- especially usage-based costing and elastic provisioning and self provisioning. More importantly most deployments as they emerge are hybrid -- both public and private, with IaaS being most prevalent in private cloud and SaaS being most prevalent in public or community cloud services.
That being said, there really are good reasons for going forward besides the bad ones like pressure, and hype, and sheer trendiness. Here is some EMA data: 76 percent of those deploying cloud services claim theyve had real, measurable dollars-and-cents benefits. Opex leads capex in cost savings by a little, but some of the other benefits may surprise you. These include improved service quality because of the resiliency that cloud infrastructures can provide applications; especially those with strong peaks and valleys in demand. Improved responsiveness to customers in terms of delivering new services when and as theyre needed is also key here and of course this is what cloud is famous for. Backup and disaster recovery also score reasonably high.
So cloud computing can be a pretty good thing.
But niche or fragmented deployments can quickly lead to saturation, or worse, breakage and vulnerabilities. Many cloud deployments, internal and external, still hover between being experimental or niche and being fully operationalized and assimilated by the broader IT organization.
Getting started well is one of the best ways to make cloud computing a strategically valuable asset, and in my next column Ill examine that.