Distinguishing cloud computing reality from cloud hype can be a daunting task. Smart Business met with Christian Teeft, vice president of Engineering at Latisys, to examine some of the key assumptions, myths and drivers that decision makers face as they build their cloud IT infrastructure strategy.
Defined: cloud computing
Cloud computing is IT infrastructure that is delivered as a service, is elastic (users consume as much or as little of a service as they want at any time), is sold on-demand and is fully managed by the provider. In short, cloud computing makes it possible to increase capacity and add capabilities as needed — without the user having to invest in new infrastructure, train new personnel or license new software. Some say the cloud is ‘[anything] as a service’ (XaaS).
The problem with this definition is that it’s very broad — little surprise there is so much confusion and noise in the cloud marketplace. So let’s drill down. There’s cloud ‘Infrastructure as a Service’ (IaaS), for businesses that outsource to data center providers and prefer to spin up and down compute resources to meet variable demand — configuring compute, memory, storage and additional services — without a dedicated physical machine (e.g. online retailers with seasonal demand). There’s cloud ‘Platform as a Service’ (PaaS), for businesses that require pre-integrated application components that they can assemble to quickly create back office systems that run and are delivered via the service provider’s platform (e.g. rich Web, mobile or social media applications). There’s also cloud ‘Software as a Service’ (SaaS), which includes the delivery of a single application through the browser to an unlimited number of customers via a set of pooled hardware resources.
We focus on IaaS as we differentiate and debunk the following assumptions.
Differentiated: private cloud vs. public
In the public cloud, apps and data run on the same shared public pool of resources — available to anyone with the swipe of a credit card. This type of environment is well-suited for certain applications such as test and development, mobility and social media, as well as some e-mail, file sharing and collaboration systems.
But when security and compliance matter, when workloads and applications are mission-critical, or when hands-on expertise is needed to tailor, migrate and optimize your environment, a private cloud solution is best.
Public cloud IaaS is cheaper than private, but you get what you pay for. Your workloads, applications and data may sit on the same server or network as a hacker from China or a spammer from Russia. You also have no visibility into the hardware your environment is running on or control over the Quality of Service (QoS) you receive. Leading private cloud providers build their platform using brand name servers, blades, firewalls and load balancers that are built for multi-tenancy and provide greater security and isolation. They also create separate resource pools so customers can choose the performance level best suited to their application. With the public cloud it’s truly ‘one size fits all.’
Debunked: cloud is best for all applications
Certain applications are clear matches for the burst-ability and elasticity of IaaS cloud: file sharing, social media, testing and development, e-mail, server virtualization and SaaS. But there are many applications where the answer isn’t as clear-cut. Legacy enterprise commercial off-the-shelf (COTS) applications require significant due diligence due to the hierarchical nature of their architecture. These applications might be a good fit for the cloud, but traditional hosting may be more practical and cost-effective. It is important to carefully evaluate the costs associated with transitioning to the cloud and be realistic about what you’re trying to achieve.
Debunked: cloud is cheaper than managed hosting
While it’s true that in the cloud you pay for only what you need to use — avoiding the need to engineer your infrastructure for peak levels of activity — the additional layer of orchestration on top of a managed hosting environment is just that: an additional software component that providers deliver to users to enable self-management, faster provisioning and granular control. This means that the same configuration consuming the same amount of resources for the same period of time is going to cost more in the cloud than it would in a dedicated environment. Cost savings are realized by leveraging dedicated hosting for predictable workloads and the cloud for variable workloads. Be sure to fully understand the ongoing usage and access fees associated with the infrastructure you are deploying to avoid any surprise charges. And challenge your vendor to tailor a hybrid dedicated and cloud solution to your specific needs.
Decision time: develop a chart
There are several important questions that must be answered as you develop your cloud IaaS strategy. Do you really need to move to the cloud? If so, which applications should be moved? And should those applications be hosted in a private cloud, with more security, reliability and support? Or are your work loads better suited for a cheaper, less regulated public cloud option? What types of templates should you start with, and who on your team will be responsible for the on-demand management of these resources?
It is important to do the research required so you can answer all of these questions, and it helps to align with a service provider that offers hands-on consultative support to right-size a hybrid dedicated and virtual solution. A provider that is equipped to assess your workload history with you and talk through potential solutions, leveraging their experience in terms of business goals, configuration development and the physical migration itself is best.
Christian Teeft is vice president of engineering at Latisys. Reach him at email@example.com.