Datasparq was engaged to help Gatwick’s car parking and commercial products team, who were looking to improve car park occupancy and pricing using advanced analytics and modern data platform capabilities.

Gatwick had previously worked extensivley to review pricing strategies and develop initial forecasts and booking curves. The team has continued to improve parking commercial strategies using data from Google Analytics to better calculate optimal pricing.

However, the data platform, in place at the time and the pricing tools demanded a lot of manual processes to deliver effective trading-decision reporting, and required a step-change in capability to drive further improvements.

Gatwick’s longer term aspiration required the new data and analytics platform to be able to support multiple data sources to deliver BI and analytics use-cases, but also be extensible to support machine learning to help with more advanced use-cases such as dynamic pricing calculations.

During the initial approach the solution was designed and deployed on AWS, making use of tools like Databricks, Microsoft SQL server and S3 storage layer that aligned with the corporate IT architecture at the time. The solution proved to be more expensive than initially anticipated and inflexible in supporting more use cases that rely heavily on SQL data exploration and machine learning.

Datasparq suggested migrating the solution to Google Cloud Platform due to its superior data analytics and AI capabilities but also to make use of the BigQuery ecosystem. The solution proved to be faster, more robust and flexible to support BI and AI applications.

Datasparq delivered a new capability to test, deploy and analyse different car park pricing strategies to improve prices offered to customers to increase revenue. This capability is now underpinned by a new GCP data platform that also increased operational trading reporting efficiencies and laid the groundwork for developing machine learning solutions.

Architecture

By following Datasparq’s engineering principles and best practices, and by leveraging our “elements of engineering” (link) we deployed a new Data Platform on GCP using serverless components and IaaC approach thourgh Terraform. The platform runs data pipelines daily having as main source Google Analytics and using Google Cloud Functions, BigQuery and PubSub in its core. The pipeline is orchestrated by our own lightweight orchestration tool called Houston which is also built on top of GCP’s tools like Kubernetes, Firestore and Memorystore.

Of course we did not stop there. To make sure the platform runs in a robust way we had to get informed about the platform's state at all times , thus we deployed our expectations testing suite called Xu (link) and made sure we are informed in real-time, through stackdriver, for potential errors.

The platform has been in Production since the summer of 2021 helping Gatwick with their daily revenue management decisions and operations.

Technologies used

• PubSub
• BigQuery
• Cloud Functions
• Cloud Storage
• Firestore for our Houston orchestration tool
• GKE for our Houston orchestration too
• Memorystore for our Houston orchestration too
• Cloud Secrets
• Stackdriver
• Cloud Scheduler
• Cloud Build
• Identity and Access Management

Superior performance and scalability.

Internal teams are benefiting from faster and more reliable performance allowing them to inform daily trading decisions.

Data lineage control.

With the current solution on Google Cloud Platform, Gatwick can track data lineage to better assess the trustworthiness of the data and avoid data misconceptions that could skew analyses.

Robust security.

Through IAM and encryption at rest and on transit, Google Cloud platform helped Datasparq deliver the robust security and governance required to access the reporting layer and the underlying datasets.