2018 is predicted to be the year of public cloud. In 24*7 IT Connection’s top technology predictions for 2018, 5 out of 9 predictions are about cloud. Turbonomic CTO, Charles Crouchman, agrees: “2018 is the year when the cloud comes into much better focus. Those that take advantage of elasticity of the IaaS layer and/or services available in the PaaS layer make the best candidates.”
Elasticity is the ability to scale resources allocated to an application up and down, vertically or horizontally, to match the current demand of resources to the supply available on the infrastructure – whether private or public. That is far easier said than done. It is very complex to truly achieve.
Elasticity comes in three flavors:
- System Infrastructure Elasticity: Increasing or decreasing the resources allocated to a single component of the application, most commonly by adding CPU or Memory.
- Software Infrastructure Elasticity: Increasing or decreasing the infrastructure the application is running on, most commonly by adding or removing a new node. AutoScaling in AWS or ScaleSets in Azure are a good example of this.
- Application Elasticity: Increasing or decreasing the resources allocated to the application, most commonly by changing heap size, cache size, number of threads and so on.
To truly be elastic you must achieve all of the above, in real-time, for all resources consumed by the application. Given that there are over 1.7 million potential considerations for EC2 instances alone and 90 additional services ontop of that, with an ever changing cost landscape and continuous expansion of the catalog, it is no surprise that most organizations consume public cloud almost statically. Gartner estimates that by embracing the complexity of the cloud and enabling automation through on going intelligent management organizations can save an average of 74% of their cloud bills.
You can dive deeper into the challenges of achieving elasticity in this blog post, but the result of this increasing complexity is that the vast majority of enterprise IT organizations don’t leverage elasticity in the cloud, or anywhere close to the full extent of its capabilities.
Turbonomic delivers automated workload optimization for hybrid cloud to solve exactly this problem, unlocking elasticity for applications running in the cloud.
Turbonomic does this by continuously, and in real time, understanding the changes in demand of resources as new transactions hit the application and models the effect of those on each component of the application all the way to the response time of these transactions. This comprehensive model represents the entire hybrid cloud estate, from the load balancers, through the different components of the application, to the provider of the application, be it containers or virtual machines.
In a public environment Turbonomic models all the possible ways the resources could be purchased across both IaaS and PaaS. This model is continuously updated as the demand of resources changes, price is updated or the available ways to purchase them is enhanced.
The Turbonomic analytics layer also constantly analyzes this model to identify what the ideal configuration would be at each layer of the stack to deliver application SLA as efficiently as possible within the constraints defined by the business.
The result is a list of trustworthy actions – changes that need to be taken on the hybrid cloud environment to bring it to a continuously healthy state. These actions could then be taken automatically by Turbonomic, manually taken by the users, or even fed into external orchestrators.
Most importantly, these actions can be taken in real time to run a hybrid cloud estate elastically to match the demand of resources to the resources allocated to the application and the best billing option for those resources. By doing that Turbonomic helps customers achieve the “E” in EC2.
 Gartner – Ten Moves to Lower Your AWS IaaS Costs – # G00326261 – Published April 2017