ArcGIS Enterprise on Kubernetes – Azure Proof of Concept

Federal Government Customer
The customer is a Federal Government entity responsible for supporting geospatial initiatives and workflows for nearly 250 million surface acres of public lands nationwide. These initiatives include mineral rights, wildland firefighting, historical lands, tribal lands, and more.

About Managed Business Solutions
Managed Business Solutions, LLC (MBS) is a full services provider of IT services and a subsidiary of Sealaska, an Alaska Native-owned Small Disadvantaged Business. With dedicated practices in Data Science & Analytics, Application Development, and the Geospatial Practice, which employs 30 highly skilled technical GIS resources, MBS provides solutions to the most challenging problems facing organizations.

The Challenge
The client maintains an ArcGIS Enterprise consisting of up to 8 Active-Active ArcGIS Server VMs in order to deliver geospatial services. The most in-demand of these services must support 8 million requests per week. Providing this level of service comes with challenges around maintaining uptime, service level agreements, reliability, and accessibility. In order to provide the highest level of support, MBS chose to evaluate ArcGIS Enterprise on Kubernetes as a potential future solution. MBS’ considerable experience with geospatial technologies, as well as with Docker and Kubernetes, made this a great opportunity to build out a proof of concept to test the capabilities of this new solution.

The Solution

Azure

Microsoft Azure
Azure is Microsoft’s cloud computing platform that provides the resources, scalability, and integrations that are ideal for deploying cloud-native applications.

Kub_50

ArcGIS Enterprise on Kubernetes
Utilizing containerization and microservices technologies to support ArcGIS Enterprise on cloud-native architecture, ArcGIS Enterprise on Kubernetes delivers the well-known environment with built-in high availability as well as greater scalability, resilience, and manageability.

terraform

Terraform
Terraform is a powerful tool for automating the provisioning of Infrastructure as Code (IaC), allowing for the definition of reliable and repeatable system deployments.

Kubernetes on Azure statistics
Load testing results from a 26+ Million record web service on the Kubernetes deployment shows response times for a high load scenario of 38k requests over 3 minutes, designed to push the system to its limits.

The Outcome
While this solution can be run in any cloud environment, MBS opted to implement this POC within our Azure stack in order to leverage the benefits of both our Esri Silver and Microsoft Gold level partnerships.

Utilizing Terraform, MBS was able to quickly and efficiently build the cloud infrastructure and deploy and configure ArcGIS Enterprise on Kubernetes in one easy step.  The results of the Terraform script is a fully functional, containerized install of ArcGIS Enterprise configured for high availability in about 45 minutes.  It is worth noting that the recommended minimum requirements produced inconsistent results, which led MBS to utilize more moderately powered resources.  For testing purposes, MBS replicated a highly utilized map service containing more than 26 million records stored in a non-indexed database.  This particular ArcGIS Enterprise install was capable of handling an extremely high load of approximately 38k requests in less than 3 minutes with stable response times.

MBS, the client, and Esri were extremely impressed with the performance results of this POC when compared to a traditional environment. ArcGIS Enterprise on Kubernetes certainly has its benefits and deserves strong consideration when architecting high-load GIS systems.

“The new Kubernetes architecture for ArcGIS Enterprise is significantly easier to deploy, configure and manage. The new cloud-native architecture, based on microservices and delivered through containerization, will help you scale to the rapidly growing demand for GIS services.”

Trevor Seaton, Product Manager - Esri