BoaviztaCloudServer

A Server whose hardware specifications and embodied carbon are looked up automatically from the Boavizta cloud-instance API. Provider and instance type pick a reference profile; the remaining parameters mirror Server.

When to use this class

Use BoaviztaCloudServer for managed cloud instances available in the Boavizta catalog (AWS, Azure, GCP, Scaleway). Use Server when you have your own hardware specifications, or GPUServer for GPU-bound workloads.

Params

name

A human readable description of the object.

provider

Cloud provider key as exposed by the Boavizta API (e.g. aws, azure, gcp, scaleway).

For example, scaleway.

instance_type

Provider-specific instance type identifier. Must be valid for the chosen BoaviztaCloudServer.provider.

For example, ent1-s.

server_type

Provisioning model of the server, which decides how many instances are attributed in each hour. Autoscaling rounds the hourly demand up to a whole number of instances, so an instance is billed even when it is only partially loaded. Serverless attributes only the fractional instance-hours actually used. On-premise holds a fixed number of physical instances over the whole modeling period (capacity sized to peak demand, or to ServerBase.fixed_nb_of_instances if set).

For example, serverless.

lifespan

Expected time before the server is replaced. Embodied carbon is amortised over this duration.

Unit: year.

idle_power

Electrical power drawn by one instance that is on but not running any jobs.

Unit: watt.

power_usage_effectiveness

Datacenter overhead multiplier applied to instance power to account for cooling, lighting, and other site-wide energy use.

Unit: dimensionless.

average_carbon_intensity

Average grid carbon intensity at the location where the server runs, used to convert energy consumption into carbon emissions.

Unit: gram / kilowatt_hour.

utilization_rate

Fraction of an instance's RAM and compute that is considered usable by jobs after operating-system and headroom overhead.

Unit: dimensionless.

base_ram_consumption

RAM consumed per instance independently of jobs, for the operating system, agents, and idle services.

Unit: gigabyte_ram.

base_compute_consumption

Compute consumed per instance independently of jobs.

Unit: cpu_core.

storage

Backing Storage attached to the server. Storage emissions are reported separately from the server's own footprint.

An instance of Storage.

fixed_nb_of_instances

On-premise only: number of physical machines deployed. Used to detect when traffic exceeds capacity. Leave empty for autoscaling and serverless server types.

User defined number of instances. Can be an EmptyExplainableObject in which case the optimum number of instances will be computed, or an ExplainableQuantity with a dimensionless value, in which case e-footprint will raise an error if the object needs more instances than available.

fraction_of_usage_time

Fraction of the modeling period during which the hardware is in active use.

Unit: dimensionless.

Fixed by BoaviztaCloudServer to 1.0 — not configurable.

Backwards links

Calculated attributes

api_call_response

Cached response from the Boavizta cloud-instance API for the chosen provider and instance type. Provides the impact and verbose hardware specification used by all subsequent updates.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

carbon_footprint_fabrication

Embodied carbon of one instance, taken from the Boavizta API response (embedded GWP impact).

Example value: 0 mg

Depends directly on:

through the following calculations:

You can also visit the link to Carbon footprint fabrication’s full calculation graph.

power

Average power drawn by one instance, taken from the Boavizta API response.

Example value: 0 mW

Depends directly on:

through the following calculations:

You can also visit the link to Power’s full calculation graph.

ram

Memory of one instance, taken from the Boavizta API response.

Example value: 0 B ram

Depends directly on:

through the following calculations:

You can also visit the link to RAM’s full calculation graph.

compute

Number of vCPU cores on one instance, taken from the Boavizta API response.

Example value: 0 cpu core

Depends directly on:

through the following calculations:

You can also visit the link to Nb cpu cores’s full calculation graph.

hour_by_hour_ram_need

Hourly RAM demand placed on the server by all of its jobs combined.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

hour_by_hour_compute_need

Hourly compute demand placed on the server by all of its jobs combined.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

occupied_ram_per_instance

RAM that is permanently occupied on each instance, summing the server's own base consumption with the base consumption of every installed service.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

occupied_compute_per_instance

Compute that is permanently occupied on each instance, summing the server's own base consumption with the base consumption of every installed service.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

available_ram_per_instance

RAM each instance has left for jobs after applying the utilization rate and subtracting RAM occupied by installed services.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

available_compute_per_instance

Compute each instance has left for jobs after applying the utilization rate and subtracting compute occupied by installed services.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

raw_nb_of_instances

Hourly number of instances strictly required to serve hourly demand, taking the maximum across the RAM and compute dimensions, before rounding to whole instances.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

nb_of_instances

Hourly number of instances actually billed, computed differently per server type: ceiled to whole instances for autoscaling, mirrored from raw demand for serverless, and held flat at peak (or the user-fixed count) for on-premise.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

instances_fabrication_footprint

Hourly fabrication-phase emissions of all instances, equal to the embodied carbon of one instance amortised over its lifespan and multiplied by the number of instances active in each hour.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

instances_energy

Hourly energy consumed by all running instances, decomposed into idle baseline energy plus the extra energy drawn while serving load, with PUE applied.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

energy_footprint

Hourly carbon emissions caused by the electricity consumed by this hardware, equal to its hourly energy use times the local grid carbon intensity.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

fabrication_impact_repartition_weight_sum

Sum of fabrication impact repartition weights, used as the denominator when normalising into per-container shares.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.

usage_impact_repartition_weight_sum

Sum of usage impact repartition weights, used as the denominator when normalising into per-container shares.

Example value: no value

Depends directly on:

through the following calculations:

You can also visit the link to no value’s full calculation graph.