Cloud Cost Optimisation (Ideas)

Yujie LiuAbout 1 min

Create a tool to detect the features of the cloud components that are used by an application. Functionalities:

Determine 'the features' by analyzing the application code or runtime
These features should be compared to the features provided by the available components.
Give advice on whether a different component could provide feature-parity at a lower cost, potentially with some trade-offs.

Research questions

We need to figure out three main things in this research:

[Cloud] What are the components that cloud service providers offer? What's the different features and prices of these components?
[Application] What kind of components are the application using? How the application use these components?
[Analysis & Advice] Based on the features offered and the features needed, design an algorithm to give advice

Benchmark the applications
- We need to figure out a set of indicators to measure the products. Common ones Like performance, availability, price. Also, the domain-specific indicators.
Distinguish common features and specific features between similar products (within/across providers).

What kind of components are the application using?

E.g. Database, Redis, Elastic Search....

Dependencies analysis (static)
- Inspect the dependencies to see that if there are any popular ones used to connect corresponding cloud components.
Network traffic analysis (runtime)
- Cooperate with some tools to analysis the protocols used by the application.

What exactly the features are the application using?

Keyword/pattern recognition in code (static)
- Eg: Database: SQL/NoSQL? Transactions? UTF-8mb4 charset? Or some features more specific that are used in the application. All these features may show some fixed patterns/keywords in the code.

Customer's concerns: some features the application used are 'must have'; but the others might be 'good to have'
The features the application using.
The features the cloud provides.
The proposed indicators of these products.