Why Self-Optimizing Apps Are the Future of Cloud

Jason Bloomberg

Jason Bloomberg IT Industry Analyst & President, Intellyx

Andrew Hillier

Andrew HillierCTO, Densify

Densify enlisted the help of Jason Bloomberg, a top industry analyst and a key Forbes contributor, and Andrew Hillier, CTO of Densify and a true veteran in the software industry, to run an informational webinar on the pitfalls of cloud application management and how to overcome them.

Trying to match application demand and public cloud supply is a major industry challenge. There are currently 3.2 million ways to procure AWS resources, and that number grows when you consider Google Cloud and Azure. Andrew and Jason discussed the fact that often, “cloud isn’t paying for what you use, but rather, what you provision.” It is imperative that companies effectively provision cloud resources when you consider the enormous amount of instance permutations.

How does a company manually and effectively manage their cloud resources? The simple answer is that they cannot.

Andrew Hillier explained that the key to cloud optimization is automation. Cloe, Densify’s Cloud Learning Optimization Engine, is at the heart of this industry solution. Cloe constantly evaluates application resource demands against the millions of permutations of cloud supply. This predictive analysis is done with the sole goal of making applications self-aware. These applications are then able to optimize themselves, removing the need for manual matching.

Watch the webinar playback

After Andrew and Jason had finished up their conversation, we conducted a question and answer segment, fielding questions on what we had covered.

How does workload pattern modelling work in regards to “bursty” workloads?
Densify models burst workloads based on a 24-hour model. However, unlike other cloud optimization companies, this modelling is tailored to time specific workloads rather than a 24-hour average. Densify’s software can visualize and identify workload peaks and separate those from applications that may have constant and sustained activity throughout the day. Based on this burst modelling, the software can recommend a certain instance that best suits the level of activity.
What does normalizing cloud catalogues mean?
Normalizing cloud catalogues is all about leveraging industry benchmarks to enable accurate comparison across instance types, and even cloud providers. Cloud providers now provide a certain level of transparency about different types and generations of instances and their underlying hardware infrastructure. One of the challenges of sizing and matching instances to application workloads is the lack of visibility when analyzing instance capacity. Densify can determine the size and capacity of specific instances relative to one another. It provides visibility and accuracy of how much bigger or smaller an instance is compared to another. This “normalizing” of cloud catalogues provides understanding to what those instances can deliver. The Densify software can then also predictively analyze what application workloads would look like on a different type of cloud node, because it has access to normalized catalogues for accurate comparison. Densify’s engine is modelling the public cloud supply based on industry benchmarks to pair specific application workloads.
Should people not implement “Infrastructure as Code”?

“Infrastructure as Code” is an important step as a maturation process for a company. However, we often find developers hardcode the values by taking their best guess of what instance type/size when deploying workloads into the cloud. This happens time and time again, and the hardcoded values are often suboptimal, either introducing operational risks or wasting OpEx dollars.

Densify enables cloud optimization to become infrastructure as code

This is not the developers’ fault as it’s simply not in their job description. More importantly, it’s not humanly possible to manually match thousands of workloads with millions of permutations of cloud resources. A machine is needed to find the perfect match, and that’s why Densify built Cloe, our Cloud-Learning Optimization Engine. This is just the first glimpse of a huge paradigm shift in the industry. Imagine if your whole cloud infrastructure optimized itself based on machine-learned behavior? Infrastructure optimization is an entire leap forward from infrastructure as code. Densify makes it easy to start this process, replacing just a single line of code to make your applications self-optimizing, and this is what we call “cloud optimization as code.