How a cloud architect predicts wildfire behavior with Google Cloud

Key Concepts:

• Google Kubernetes Engine (GKE): A managed Kubernetes service provided by Google Cloud.
• Ember Spread Model: A wildfire spread prediction model.
• Deployments: Kubernetes objects that manage replicated applications.
• Jobs: Kubernetes objects that run finite tasks to completion.
• Pods: The smallest deployable units in Kubernetes, containing one or more containers.
• Firestore: A NoSQL document database offered by Google Cloud.
• Cloud Storage: Google Cloud's object storage service.
• BigQuery: Google's fully-managed, serverless data warehouse.
• API Requests: Requests made to an application programming interface.

Wildfire Spread Model on GKE

The speaker, originally from Bolivia and a computer science graduate, was hired by Improving Navigation and worked with the wildfire team using the Ember spread model. A core responsibility was creating the entire cluster on GKE.

GKE Cluster Configuration

The speaker was responsible for configuring the GKE cluster, including:

• Selecting appropriate compute resources.
• Determining the number of nodes.
• Deciding whether to use Deployments or Jobs.
• Optimizing memory allocation.

The speaker notes that GKE's monitoring and logging capabilities simplified this process.

Key Components of the Wildfire Model

The wildfire model has two main components:

1. Constant Deployment: A Deployment that is always running and listening for API requests.
2. Job Creation: Upon receiving an API request, the Deployment creates Jobs on GKE. These Jobs are automatically deleted after 10 minutes once they complete their task.

The Ember spread model runs inside the Pods created by these Jobs.

Data Storage and Analysis

The system utilizes several Google Cloud services for data management:

• Firestore: Stores metadata related to the model runs, including fire parameters and Ember spreads.
• Cloud Storage: Stores data in buckets.
• BigQuery: Receives data for analysis.

Security

The speaker highlights the robust security features of Google Cloud Platform, emphasizing the need for proper authentication and setup to ensure a secure environment.

GKE Experience

The speaker notes that it was their first time working with GKE and found the learning process to be smooth due to the extensive documentation and guides available.

Conclusion

The speaker successfully implemented a wildfire spread model on GKE, leveraging various Google Cloud services for compute, storage, and data analysis. The speaker emphasizes the importance of GKE's monitoring capabilities, the security features of Google Cloud, and the availability of comprehensive documentation for a positive development experience.