On Demand

On Demand Compute

Spheron offers On Demand compute to adjust the computational resources allocated to an instance. Users can make decisions to increase or decrease the allocated resources, such as CPU, memory, or storage, to meet changing performance requirements. This type provides users with direct control over resource adjustments, allowing for optimization based on specific workload characteristics and immediate operational needs.

⚠️

The On Demand compute is currently in its closed alpha stage. Potential irregularities may lead to brief service downtimes. We recommend utilizing this feature exclusively in testing or development environments until its official release for optimal stability. If you're interested in obtaining alpha access, please use this link (opens in a new tab) to schedule a call with our team.

Get Alpha Access

Please schedule a team call (opens in a new tab) to gain early access to On Demand Type.

How to use On Demand Compute?

With Docker

To use On Demand compute with a custom Docker image on Spheron:

  1. Click "New Cluster" on the top right corner.
  2. Select Import from Docker Hub.
  3. Enter the names for your cluster and docker image.
  4. Then, Add the tag and Click "Next".
  5. Select "On Demand" under Compute Type.
  6. Select your preferred Region, if any. If you do not add a region, the container will be deployed in the eu-east region. Click here to know more.
  7. Select the instance plan that suits your needs. Use the "Create Custom Plan" toggle to create custom plans for your instance.
  8. Configure Storage (SSD) plan for your instance. Use the "Add Persistent Storage" toggle to add persistent storage for your instance.
  9. Create new Port Policy Mapping. Add the container port, and Select the exposed port you want to map it to. Click here to know more.
  10. Add Environment Variable, if any.
  11. Add Secret Environment Variable if the value is a secret key. It will not be saved in the database. Click here to know more.
  12. You can add advanced configuration if required. Click here to know more.
  13. You can add health checkup if required. Click here to know more.
  14. Click "Deploy" to initiate deployment.

With Marketplace App

To use On Demand compute with a marketplace app on Spheron:

  1. Click "New Cluster" on the top right corner.
  2. Select Start from marketplace app.
  3. Pick your desired template from the marketplace.
  4. Select "On Demand" under Compute Type.
  5. Select your preferred Region, if any. If you do not add a region, the container will be deployed in the eu-east region. Click here to know more.
  6. Select the instance plan that suits your needs. Use the "Create Custom Plan" toggle to create custom plans for your instance.
  7. Configure Storage (SSD) plan for your instance. Use the "Add Persistent Storage" toggle to add persistent storage for your instance.
  8. You can add advanced configuration if required. Click here to know more.
  9. Click "Deploy" to initiate deployment.

How to update configuration of your instance?

To update the configuration of your instance:

  1. Select your instance and Go to the Settings tab.
  2. Click "Update Instance" under the Instance Plan section.
  3. Update the instance plan to suit your needs. Use the "Create Custom Plan" toggle to create custom plans for your instance.
  4. Configure Storage (SSD) plan for your instance. Use the "Add Persistent Storage" toggle to add persistent storage for your instance.
  5. You can update advanced configuration if required. Click here to know more.
  6. Click "Update" to initiate deployment.

Limitations

  • Human Intervention: On Demand compute requires human intervention, which can lead to delays in responding to changing workloads.
  • Scalability: On Demand compute might not be efficient for systems that need to handle varying workloads. It can be challenging to predict the exact resource requirements for different levels of demand, which can result in under-provisioning or over-provisioning of resources.
  • Downtime and Disruption: On Demand compute resources manually can lead to downtime or disruption in service. Adjusting resources may require stopping or restarting parts of the system, leading to potential service interruptions for users.
  • Complexity: As the system grows, the complexity of manually managing resources increases. In large and distributed environments, keeping track of resource allocation and utilization becomes more challenging, resulting in poor performance and user dissatisfaction.
SpotAutoscale