Cloud Computing is the future of corporates. Due to this pandemic, as more and more people are forced to move to Cloud from their traditional on-premises solutions, there has been a rise in Cloud Service Providers. These providers offer pay as you go model and hence cuts off the Cap-ex costs. When an organization moves to the Cloud, the collection of networked elements delivering services ought not to be independently handled or maintained by users. Instead, the whole provider-managed suite of software and hardware can be thought of as an indistinct cloud.
What is Custom Cloud?
Custom Cloud is a typical selection of cloud services for optimizing cloud services depends on a number of conflicting quality of service measures. In theory, we deliver a customized cloud in which users get the highest quality of service after comparing with the available cloud services based in their application and needs, based on various functional and non-functional service characteristics, with the aid of multi-corder preference.
How does Custom Cloud make a difference?
It is user-centric and compares various parameters and attributes like Response time, Availability, Throughput, Successabilty, Reliability, Compliance, Best Practices according to WS-I Basic Profile, Latency, Documentation, and Web Service Relevancy Function. Due to this pandemic, most of the people were not able to access their physical systems and will want to shift to Cloud for their computational and storage needs. We will provide the custom made options which best suits their needs, and this will be one of the Business aspects of the project. After we have gained a certain number of users and provided them with solutions and gained Market Trust, we plan to approach Cloud Service Providers.
How does Custom Cloud work?
Following steps are followed in the Custom Cloud backend when a User enters an input file:
- User Requirements — Needs of users are taken as in input in particular. Their preference is then passed on to the next step.
- Research and Analysis — After the research, I learned about existing solutions to the same problem of CSP selection and different MCDM techniques to compare services.
- Design — Post research and analysis, I proposed my own algorithm. Implementation — Implementing the proposed algorithm as well as the top existing algorithm. If it fails, go back to the Design step.
- Comparison — Training the proposed algorithm on Neural Network and comparing results with both Proposed and top existing algorithms to minimize mean square error. If it fails, go back to the Design step.
- Testing — Test the algorithm on the user’s data and reduce time and space complexity. If a problem occurs, go back to the implementation part.
The input file is based on user preferences where the users can prioritize attributes over another. Following is an example of an input file that has to upload to get the best Cloud Service Provider:
To test and showcase the capstone project, I created a prototype — a simple to use UI where the user needs to upload the above-shown input file with all his preferences and priorities and after a comparison based on 10 attributes, he gets the best CSP as an output based on his particular choices and preferences. There are basically two steps, where the user can download the sample preference file, update the preferences, and then upload the updated file in step 2 to check the best Cloud Service Provider based on your preferences. Following is a screenshot of the prototype User Interface:
Why is Custom Cloud better?
Custom Cloud uses a combination of different MCDM methods like AHP, TOPSIS, ANP, and VIKOR to train the Neural Network which then predicts the best Cloud Service Provider based on user preferences, hence reducing the time complexity for faster and more accurate results. You can now compare from over 2500+ CSPs within a split second!
WebApp — Experience Custom Cloud on the WebApp.
GitHub — Find the source code and all the details on my Git repository.
YouTube — See the youtube video on Custom Cloud.
I hope you liked the read. Please write back at bothera.abhi[at]gmail.com in case of any queries and feedbacks/suggestions.