Data-driven resource management and the future of cloud


Cloud adoption is undoubtedly the cornerstone of digital transformation, and for many, it is the foundation for rapid, scalable application development and delivery. Companies of all sizes and from across all industries are racing to achieve the many benefits afforded by public, private or hybrid cloud infrastructure. According to a recent study, 20 percent of enterprises plan to more than double public cloud spending in 2018, and 71 percent will grow public cloud spending more than 20 percent.

Enterprises moving to the cloud are often seeking to improve employee collaboration, ensure redundancy, boost security and increase agility in application development. One of the top advantages afforded by the cloud is the ability to auto-scale in response to demand — a feature that has transformed what was once capacity planning into a more continuous cycle of capacity and resource management.

The impact of the cloud

When it comes to physical data centers, capacity planning primarily entails predicting, purchasing and installing the maximum number of servers an organization may need. As traffic needs change over time, data centers can be expanded or consolidated, but changes are slow and cumbersome. Capacity must be able to accommodate brief spikes in traffic, otherwise performance and uptime are negatively impacted. Software engineers must be, in effect, “magicians” who use predictive modeling matching application needs and traffic drivers to resource constraints in order to determine the correct capacity. While planning tools can be beneficial, the complexity of this setup causes most to take the “wait until it breaks” approach to determine what the upper limits are.

With cloud computing, a burst of traffic can be more easily addressed as modern enterprises can quickly spin up new services and capacity to dramatically improve user experiences. This flexibility allows organizations to account for both expected spikes or unexpected conditions, such as the historically well-known “Slashdot” effect. Whether it is predictable or not, we can now build applications to respond to events because of the automation and flexibility enabled by the cloud. This idea of “infinite” capacity and elastic infrastructure is appealing, and devops processes and tools like Terraform have improved the speed at which scale is possible. At NS1, for example, our team has built what is essentially a push-button operation to stand up a new cloud deployment of our entire platform, from scratch. This process — which previously would have taken weeks or months — has been reduced to a matter of minutes.

Leave a Reply

Your email address will not be published. Required fields are marked *