Skip to main content

Deployment Overview

Welcome to the evoML deployment documentation. This guide outlines the supported deployment options for evoML and provides essential information to ensure a smooth and successful implementation.

evoML offers two primary deployment options:

  • Kubernetes-based Deployment: Leverages Kubernetes for container orchestration and scalability. Recommended for environments requiring high availability and resource management.
  • Docker-based Deployment: Provides a simplified deployment approach using Docker containers. Suitable for smaller-scale deployments or development environments.

Select the deployment option that best aligns with your infrastructure requirements and operational considerations.

Both deployment methods support:

  • Standard Installation (Internet Access): Utilizes standard package management and network connectivity.
  • Air-Gapped Installation (Secure Environments): Designed for environments with limited or no internet access, ensuring secure deployment. See the dedicated "Air-Gapped Installation" section for specific instructions.

Pre-Deployment Information Gathering

To facilitate a seamless deployment experience and ensure evoML is optimally configured for your environment, please provide answers to the following pre-deployment questions. This information will help us tailor the installation process and provide appropriate support.

Please submit your responses to your dedicated TurinTech contact point or email them to support@turintech.ai.

Important: Please provide your answers to your dedicated TurinTech contact point or email them to support@turintech.ai.

  1. User Concurrency: What is the anticipated number of concurrent users who will be actively using the platform?

  2. Concurrent Trials: How many concurrent trials (model training and evaluation processes) are expected to run simultaneously? This will inform resource allocation.

  3. Deployment Location: Where will the evoML instance be deployed? (e.g., AWS, Azure, GCP, On-Premise) Be specific as possible.

  4. Access Permissions: What level of access permissions can be granted for the deployment process? Please specify one of the following options:

    • Kubernetes Cluster Access Only: Limited access to the Kubernetes cluster for deployment purposes.
    • IAM User: An Identity and Access Management (IAM) user with specific permissions for cloud resource management. Please specify the IAM policy.
    • SSH Access: Secure Shell (SSH) access to the underlying infrastructure (e.g., bare-metal servers). Required for bare-metal deployments.

  5. Domain Integration: Will evoML/Artemis be deployed within your existing domain?

    • If yes, please specify the domain registrar and location (e.g., Route 53 in AWS).
    • Are there any restrictions on network access to the domain? (e.g., limited to specific IPs or an internal network) If so, please provide details.

  6. Data Sources: What types of data will be used with evoML, and what are the sources of this data? (e.g., CSV files, databases, APIs)

  7. Estimated Cloud Costs: What is the estimated monthly cost for the default proposed deployment on your cloud provider of choice (AWS, Azure, GCP)? This helps us validate the proposed configuration.

  8. Infrastructure Maintenance: Will the Kubernetes cluster or the underlying machines used to deploy evoML be restarted or undergo regular maintenance? If so, what is the typical frequency of these events? Understanding the maintenance schedule is crucial for ensuring service availability and planning for potential downtime.