Compute Component
The compute component is of core importance and all platforms since it represents the virtual machine and operating system on which the platform runs.
You can configure the compute component as part of your platform in design phase and specific to an environment in the transition phase.
Once your assembly is deployed in an environment you can access the computes in operation.
Configuration
Besides the global configuration available for any component such as Name and Description, you can configure the following attributes:
Instance Size: The instance size determines characteristics of the virtual machine created for operation in terms of processing power, memory size, networking bandwidth and operating system. The size values use clothing sizing values of from extra small to extra large and beyond - XS, S, M, L, XL, XXL, 3XL, 4XL. Instance sizes optimized for compute performance, network performance, storage and memory are available. The generic values are mapped to cloud specific sizes.
Networking - PAT ports: Configure the Port Address Translation PAT from
internal ports (key) to external ports (value).
Networking - Require public IP: Check if a public IP is required. Setting is
used when the compute cloud service public networking type is interface or
floating.
The Cloud Services configuration displays the services required by the component and provided by the cloud. Typically compute and dns are required, while others such as mirrors or ntp are optional and can be enabled or disabled as desired.
The Compute Depends On and Depend On Compute sections contain lists of related components.
The attachments tab should not be used on computes components as attachments are not functional when associated to a compute.
The monitors tab can be used to configure compute-related monitors.
Example Use Cases
Update the Size or OS of a Compute
Changing a compute in design, like any other design change, requires you to:
- Save the change and commit the overall design.
- Pull the design to the environment.
- Deploy the environment.
If you are changing a compute configuration like size or a related setting all deployed instances have to be flagged to be replaced.
To roll out a change you need to either disable and re-enable the whole platform perform a rolling replacment.
A platform wide approach means that the application will be unavailable during the procedure.
- Change the configuration of the compute in design.
- Set the action to replace all the computes in operation.
- Disable the entire platform.
- Commit and deploy.
- Enable the platform to commit and deploy.
Alternatively you can roll the change out via replacing computes:
- Change the configuration of the compute in design.
- Set the action to replace all the computes in operation.
- Choose a step size of less than 100% for a rolling upgrade.
- Pull the design changes to the environment.
- Deploy to the environment.
