What to Consider When Moving to the Cloud

Multiple cloud pricing models are available in the market today. Within each category of offerings detailed in this section, most cloud service providers offer a multiplex of offerings that consider the currency of the underlying CPU, the number of virtual cores or virtual CPUs, the amount of RAM and storage required, and the associated payment terms. Customers must first select the general approach they want to use and then drill down deeply into the exact configuration that will meet their requirements.

For most customers today, a mix of infrastructure-as-a-service (IaaS), platform-as-a-service (PaaS), and software-as-a-service (SaaS) resources will serve their needs. For large enterprises with a substantial investment in traditional datacenters, the need for IaaS and PaaS solutions helps drive their economic decisions. This IDC White Paper focuses primarily on IaaS consumption, but customers desiring PaaS resources will find compelling pricing in that dimension as well, with similar pricing models in use for container-based and other PaaS environments.

Most cloud providers offer customers and prospective customers limited-length, limited-capacity free access to cloud resources, so customers can deploy an experimental application without any up-front financial commitment. This makes it easier for developers or administrators to try out a cloud environment without engaging management or needing to secure a purchase order.

Development/test instances used by developers generally include more resources and longer use periods than no-cost trial instances. These developer instances sometimes are a low- or no-cost benefit of a larger overall relationship with a given cloud vendor. In some cases, cloud providers offer accompanying benefits such as discounts (trial or longer term) on related products or free access to other software, development tools, and other cloud resources and services.

Pay as you go is no-commitment billing for use of cloud computing services. Users typically pay only for the resources they consume, but the rate they are paying per time unit tends to be substantially higher than that of other subscription types, such as reserved instances. Pay as you go is a cost-effective strategy that requires no up-front investment.

A reserved instance is a commitment for a specific period for a given workload or service. The rate per time unit tends to be considerably lower than that of pay-asyou-go services, but a reserved instance is expected to be billed for 24 x 7. It is effectively an analog to having a server running around the clock in a customer’s datacenter. In cases where customers have a steady state application that needs to be “always on,” such as a database, this can be the best approach, although an application need not be operated around the clock to be more cost effective than pay as you go.

Customers can subscribe to a dedicated server (host) that is theirs to use for a predetermined period of hours or perpetually as a reserved resource. The benefit of using a dedicated host is that customers having software licenses that are tied to a server can consume that resource because the licenses will be deployed on a known host with a known configuration (number of processors, number of cores, memory, etc.) that may meet the software licensing terms and conditions

This approach make sense from a software licensing perspective but loses some of the prime benefits that cloud computing brings to customers in terms of agility and ability to purchase only the computing resources they need. Customers have dedicated use of that server, meaning if they are underutilizing the server, there might be no way to rebalance that workload with the capacity it requires.

Most cloud service providers offer some form of on-demand capacity boosts and use different terminology to describe this solution. In some cases, there are offers that include virtual machines (VMs) that are, by definition, lower capacity and able to burst to additional VMs when required. Other cloud vendors offer other unique approaches to acquiring and disposing of ad hoc capacity.

The opposite scenario from ad hoc capacity boosting is the release of unused reserved instance capacity. As with ad hoc capacity boosting, there is no standard approach to releasing capacity. Some clouds offer a guaranteed return value, while others require users to sell capacity back to another customer. In both boosting and returning of ad hoc capacity, it behooves customers to consider how likely they will be to need to acquire or dispose of additional resources.

Some cloud providers offer access to spare capacity that can exist in each region, for a significantly reduced price, for off-hours computing needs. Optimizing consumption of this spare capacity can significantly reduce costs, especially where there is less urgency to complete a job. Applications where reduction of cost is important and applications that can tolerate nodes being terminated fit well here. Some high-performance compute tasks could fit in this category.

Another dimension that is inconsistent in the industry is the ability to bring your own license to cloud. The combination of independent software vendors and cloud service providers may impact the options that are available. If they are one and the same, portability options may be offered that are not available for a multivendor scenario. In some cases, customers are required to reserve an entire server to be able to reuse an existing license for a cloud deployment. In other cases, there may be options to migrate a license from on-premises to a cloud deployment scenario, allowing customers to acquire cloud computing resources at a lower cost.