Explore All Case Studies.
Firstsource Solutions Limited is a leading provider of customized Business Process Management (BPM) services. Firstsource specializes in helping customers stay ahead of the curve through transformational solutions in order to re-imagine business processes and deliver increased eﬃciency, deeper insights, and superior outcomes. They are trusted custodians and long-term partners to 100+ leading brands with presence in the US, UK, Philippines, and India.
Built an Enterprise-level Solution to Predict the Call Volumes Using Kubernetes & GCP
25 Lakh +
Challenges and Problem Statement
Firstsource wanted to leverage the data collected across various business verticals like Retail, BFSI, telecom, and bring forth additional insights for their clients using Analytics. As their codebase was growing in size, they wanted to modernize the architecture of their Analytics platform, especially in terms of portability and scalability, which included an efficient way to deploy new features and modify old features. They also wanted to contrive a highly scalable and fully secure way to integrate the vast amount of data coming from various sources.
They have a strong data science team to build the data models for building a prediction model. Niveus proposed to collaborate with them to enhance the existing Analytics platform to a scalable SAAS offering, which is easily deployable and secure given they were dealing with a lot of customer data.
They were looking to enhance their existing architecture & make it more scalable while further bringing down the cost. The Microservice-based architecture appealed to them and they decided to go for a change!
Firstsource preferred the microservices-based architecture because of the extended cloud storage that can be used for extensively used node or service. Their objective to have an easier-to-maintain and easier-to-integrate analytics platform were amongst the other reasons that drove them towards choosing containerization.
Objective was to build a prediction system that will forecast the call volumes for the near future, based on historical data. Building a multi-tenant SAAS platform to serve customers with multiple machine learning data model across industry vertices, was the best solution. To achieve the desired objective, an integrated platform that has integrations with various systems and option to import various source data has been built using microservices-based architecture where all components have been converted to microservices using wrappers and run on Google Cloud Platform (GCP) using Google Kubernetes Engine (GKE).
GKE has a whole slew of advantages: automated orchestration, deployment, scaling of containers, reducing the learning curve significantly and supporting the latest versions of Kubernetes sooner, to name a few.
The key tenets for multi-tenant SAAS platform design were:
- A platform to cater multiple customers with pick and choose data model
- To allow for easy build and deploy pipeline.
- To ensure that the smallest piece of fix can be easily deployed by the developer using microservices-based architecture
- To provide an environment to build, deploy and test
- To de-risk any releases to production & ensure 100% availability of the platform utilizing the Canary model
- Easy scalability based on the computing load and customers
A set up has been established with an option to define the parameters, change and tweak values, override outliers, and use system suggested replacements. Now, with the enhanced architecture, the system will automatically choose the best algorithm for prediction, based on previous accuracies. The newly created Dashboard will help them visualize and analyze prediction results and compare with the previous runs.
We effectively leveraged the power of the Google Cloud Platform and Kubernetes to enhance their existing systems’ scalability and performance. This provided a configuration-driven framework where the company can get the entire network, disk, and application spun up and with just one click or set up an automated scaling based on traffic and various parameters, thus making it more dynamic in nature and easy to manage. Other than automatically bootstrapping the cluster, the combination of GKE and Kubernetes eased the management of the cluster and made it more seamless to use in the long run.
- Microservice-based Containerized modern integration system
- An improvement of 20% in the prediction time
- Ability to perform multiple prediction runs with improved accuracy, simultaneously
- Enhanced security by integrating additional security features – Cloud Security Command Center, Forseti Security, and Binary Authorization.
- The highly scalable microservice-based architecture and containerization has provided them an improved platform in terms of deployability, adaptability, ease of use and ease of integration.
- Moving to microservice-based architecture using Kubernetes gave the customer their desired scalability, where they can scale components independently.
- Agility with the decentralization of data management using microservices.
- Faster time to market for new feature releases
- Easy integration and automatic deployment
- Leveraged the power of cloud security to enjoy a more secure architecture
- A cost-efficient solution that is easy to use.
“I wanted to thank you and the team for the hard work they have put in developing the AnalyticsFIRST platform. The microservices architecture and the approach of leveraging Kubernetes cluster on GCP to containerize the platform gives us the flexibility right from development to deployment. The groundwork done by the team for integrating various Analytical API’s has helped us in making it an enterprise-grade platform. It is worth noting the extra effort that the team put in to make sure that the platform passed all security tests and ready to ingest, store and work with various financial and healthcare datasets. I hope this is just the start as we embark on our marketing and sales journey for the AnalyticsFIRST platform. Thanks and Keep up the good work!!”