Federal Budget Considerations for FY2016 Cloud Acquisition Previous item Your customers are at... Next item Some Cloud Services are...

Adoption of cloud technologies by federal agencies continues to accelerate. There have been several high-profile contract awards granted in 2015, including a $109M FAA Cloud Contract and several large GWACs, such as SEWP V, providing a readily-accessible contract vehicle for procuring a broad selection of technologies including commercial cloud services. As the Federal 2015 fiscal year draws to a close and agency technologists, procurement staff, and contracting personnel begin to plan for the FY2016, there are several strategies that can enable agencies to deploy more flexible and agile solutions while providing for budget certainty and predictability.

Indeed, one of the primary concerns about the automated scaling, on-demand, and self-provisioning aspects of commercial cloud services is the ability to accurately predict cloud spending while staying within allocated budget. In addition, while many agencies are identifying workloads for migration to commercial clouds, they simultaneously face the reality that many workloads cannot be migrated to a commercial cloud environment. This often includes workloads that are deployed on legacy infrastructure and/or workloads that are subject to higher security impact levels.

Therefore, agencies must consider how to best leverage the benefits of the cloud, while also accommodating workloads that cannot be migrated to the cloud, and stay within budget. Below are some things that agencies should consider as they navigate this landscape throughout FY2016:

1. “On-Demand” vs “Reserved”: On-demand cloud resources are great for test/dev, but are often not ideal for production workloads. Consider cloud providers that offer “Reservation Pools” of dedicated resources. Reservation pools typically provide a defined amount of cloud resources at a fixed monthly price which makes budgeting far easier. With proper planning, reservation pools can be statistically oversubscribed without impacting overall performance, which can provide better economies-of-scale and better utilization rates.
2. Beware of ALL Cloud Costs :Many cloud providers allow for dynamic, automated scaling of virtual machines based upon demand. While the ability to automatically scale is a great technology feature, it is also a potential source of additional cost which may not be budgeted. In addition, beware of fees assessed for data transfer. Data transfer fees can be difficult to predict and make removing data from the cloud more expensive.
3. Consider “Owning” Your Steady-State : Cloud economics can change significantly as utilization increases. In many cases, agencies that can predict the steady-state utilization of their deployed systems can realize better overall economics by owning their “steady-state” infrastructure and bursting their peak workloads into a commercial cloud environment. A strategy of owning the “base” and renting the “peak” provides better long-term economics while simultaneously accommodating utilization spikes.
4. Choose the Right Technology Partner : Consider a technology partner that can provide commercial cloud services and host private infrastructure. For those workloads that are not “cloud-able,” consider working with providers that can accommodate those workloads on dedicated private infrastructure within a compliant, commercial data center. By working with a service provider that can provide a broad set of service offerings, including public cloud, private cloud, and data center colocation under a single roof and compliance office, agencies can target the right solution for each workload using a single vendor.

This provides for easier contracting, and can ease migration of workloads as their technology architecture evolves over the next several years.

Want to learn more on how your organization can assist agencies to meet Federal “Cloud-First” mandates? Join Comstor and QTS on Tuesday, October 20 for an educational webinar about cloud trends. Click here to register.