The client is a major European carrier. Established in 1928, it’s one of the world's oldest operating airlines. With a fleet of almost 100 aircrafts, the Client flies to over 120 destinations across Europe, Asia and North America.
|Technology||Azure, Synapse, Databricks, Machine Learning, MLOps, Kubernetes|
- Enhanced analytics capabilities thanks to an extendible data management platform
- Robust architecture that supports the integration between the cloud and systems located on-premises
- Cost optimisation resulting from the scalability of the solution
The Client, a major airline company, strives to provide the best service to millions of passengers every day. In order to improve various processes, they were putting together specialised teams of data scientists who would use advanced technology to assist the business in achieving this goal.
The data scientists were working with the company’s complex landscape of systems, which had been growing over time. Its size and complexity didn’t allow the teams to timely access the data they needed. The Client’s technology landscape was very broad. This was preventing the teams from quickly validating their ideas and deciding if they should proceed with a certain project. They were missing tools that would allow them to make informed decisions as swiftly as they needed. The Client understood that cloud solutions could be an answer to this problem. However, they were unsure if a rapid migration from their current landscape to a cloud-based one would fit within their business objectives and cost strategy.
The Client was already using Microsoft services and reached out to the cloud provider in the hope that they would suggest the right course of action. Microsoft nominated Objectivity as a highly skilled technology partner. As a Microsoft Gold Partner with the Advanced Specialization in Modernization of Web Applications to Microsoft Azure, Objectivity could be entrusted to provide the Client with the answers they were looking for.
The Objectivity team was tasked with delivering a Proof of Concept (PoC) to show how cloud technology (applied according to the Client’s needs) could address the company’s analytical challenges. The PoC would also prove that, with Azure Cloud tools, the Client’s teams would be able to quickly access various types of data and leverage data science to solve major business issues.
It was necessary to include multiple functionalities in the PoC and make sure that the architecture was flexible, allowed for quick introduction of new types of data sources, and could be further developed in the final solution. The first element of the project was the creation of an Azure Landing Zone as the development environment. At the same time, the Objectivity team designed the architecture and analysed the business case.
The team performed an analysis of Big Data from an extensive data source — over 300 variables from plain sensors, recorded at a 5 second frequency during each flight. Objectivity used the power of Databricks to collect, process and aggregate the data which was then uploaded to a data warehouse in Synapse. The Objectivity team also processed data from other sources (e.g., CSV files, on-prem databases, XML data). Previously, Power BI was used as the data model and transformation layer, and its load was now transferred onto a data warehouse that was easier to manage and scale out. This allowed for the data to be processed in a single place and become available to different user groups. The integration process was coherent and well-monitored thanks to the use of Azure Data Factory.
Objectivity trained a flight delay ML model that would be able to determine the probability of flight delays and their extent. The model was then deployed to Azure Kubernetes Services. The new MLOps process allowed the model to continue to learn and produce predictions in the Client’s areas of interest.
The architecture built by the Objectivity team was also complemented with various new components e.g., bots, streaming, and live dashboards. The team processed a portion of the Client’s data in the PoC, however, the architecture they created can be easily scaled to cover giga- or terabytes of data.
Upon the completion of the PoC, the Client received a robust solution that was ready to be deployed to production at any time. Moreover, the Client was provided with an accurate forecast for the next 3 years and is now able to continuously expand their new system. The architecture designed by Objectivity supports the integration between the cloud and on-prem solutions. The Client can continue to assess their analytics needs and add new functionalities to their solution at a pace that fits within their business dynamics. The PoC was supplemented with a detailed cost analysis which provided the Client with the validation they needed to plan the expansion of the solution.
The PoC presented the Client with a convincing argument for a revision of their data integration landscape and for granting their teams easier access to the data they need. The Client’s teams now have tools that allow for quick validation of a hypothesis and don’t generate costs when they’re not being used. The Databricks and other functionalities can be turned off or on, depending on the Client’s needs.
The Client’s teams were trained to fully exploit the capabilities of the PoC and empowered with knowledge that enables them to develop the solution further without external guidance.