-
Notifications
You must be signed in to change notification settings - Fork 241
feat(efc): add instances metho MTA-5653 #4971
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
base: main
Are you sure you want to change the base?
Changes from all commits
19d96a5
c8bb646
7ff4bd2
1453b84
ee04787
663eee4
2d281fd
f26b10d
5d6fa98
61595e1
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,41 @@ | ||
--- | ||
meta: | ||
title: Bare Metal footprint calculation | ||
description: This page explains how Scaleway calculates the environmental footprint of its Bare Metal servers in detail. | ||
content: | ||
h1: Bare Metal footprint calculation | ||
paragraph: This page explains how Scaleway calculates the environmental footprint of its Bare Metal servers in detail. | ||
tags: bare-metal environmental-footprint impact datacentre | ||
dates: | ||
validation: 2025-05-27 | ||
posted: 2025-05-27 | ||
categories: | ||
- environmental-footprint | ||
- bare-metal | ||
--- | ||
|
||
<Message type="note"> | ||
This page includes the specificities of the environmental footprint calculation for Bare Metal servers, which also takes into consideration all the elements described in the [Environmental Footprint calculation breakdown](/environmental-footprint/additional-content/environmental-footprint-calculator/) documentation page. Refer to said page for a full breakdown of the Environmental Footprint calculation performed at Scaleway. | ||
</Message> | ||
|
||
For bare metal servers, the allocation of environmental impact is straightforward due to the direct association between a user and a server. The environmental impact of a bare metal server is entirely attributed to the user who operates it. This encompasses the energy consumption during its operation, as well as a portion of the impact from the server's manufacturing process. | ||
|
||
The manufacturing impact is prorated based on the server's lifespan and the duration of its use by the customer. By dividing the total manufacturing impact by the expected lifetime of the server and then allocating it according to the specific usage period, we ensure that each user is accountable for their fair share of the server's overall environmental footprint. | ||
|
||
<Lightbox src="scaleway-allocation-rules-customer-it-equip.webp" alt="" /> | ||
|
||
<Message type="important"> | ||
Currently, our calculations for Elastic Metal GPU servers do not include the environmental impact of GPU manufacturing due to insufficient data from the manufacturer. While the impact of CPUs, RAM, disk, and energy consumption is accounted for, the absence of GPU manufacturing data means the figures provided today are underestimated. | ||
|
||
We will update our methodology as soon as more reliable data on GPU manufacturing becomes available. | ||
</Message> | ||
|
||
## Calculation example | ||
|
||
For the use of an Elastic Metal server, with a lifespan of 6 years (52 560 hours) and power of 110w, for 100 hours and considering via Boavizta estimates that the impact of the manufacturing of this server is 850 kgCo2e, we make the following calculation: | ||
|
||
```math | ||
ServerManufacturing = (100 / 52560) * 850 = 1,62 kgCO2e | ||
``` | ||
|
||
Therefore 1,62 kgCO2e is added to the total manufacturing impact for the use of this server. | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. This example explain how the manufacturing impact is calculated but not the use. For the use, we will have : So for this exemple with an Elastic Metal with a power of 110w, use for 100 hour in DC5 (french energy mix = O,056kgCo2e/kwh and dc PUE = 1,16) we will have : |
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,98 @@ | ||
--- | ||
meta: | ||
title: Instances footprint calculation | ||
description: This page explains how Scaleway calculates the environmental footprint of its Instances in detail. | ||
content: | ||
h1: Instances footprint calculation | ||
paragraph: This page explains how Scaleway calculates the environmental footprint of its Instances in detail. | ||
tags: instances environmental-footprint impact cloud-computing | ||
dates: | ||
validation: 2025-05-27 | ||
posted: 2025-05-27 | ||
categories: | ||
- environmental-footprint | ||
- instances | ||
--- | ||
|
||
For Virtual Instances, the calculation of the environmental footprint takes into consideration all the elements described on the [Environmental Footprint calculation breakdown](/environmental-footprint/additional-content/environmental-footprint-calculator/) documentation page. | ||
|
||
|
||
## Calculation aspects | ||
|
||
The following elements are considered in the Instance footprint calculation: | ||
- **Hypervisor resources** - the resources (CPU, RAM, and disk, for example) used in the physical hypervisor servers that create and run the virtual Instances. | ||
- **Instance offer resources** - the vCPU, RAM, and disk resources used when you use an Instance. | ||
|
||
The calculation using the elements above can be broken down into: | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. We can add this to the bare metal server page too. It's a good summary of environmental footprint calculation breakdown. |
||
|
||
### Manufacturing Impact | ||
- Manufacturing of the underlying physical servers (hypervisors) necessary to run the Instances. | ||
- Distribution of the manufacturing impact according to the resources consumed by the Instance (vCPU, RAM, storage). | ||
|
||
### Operational Impact | ||
- Energy consumption of the hypervisors during Instance usage. | ||
- [Power Usage Effectiveness (PUE)](/environmental-footprint/concepts#pue-power-usage-effectiveness) specific to the data center where the Instance is hosted. | ||
- Energy mix of the country that provides the electricity. | ||
|
||
### Usage Impact | ||
- Resources consumed by the Instance, calculated based on the material specifications of the hypervisors. | ||
|
||
### Indirect Emissions | ||
- Impact related to cross-functional services necessary to guarantee the operation of the Instance, such as network and shared storage. | ||
|
||
## Instance consumption ratio | ||
|
||
When calculating an Instance's footprint, we consider the resources based on their relative usage on the hypervisor to ensure an accurate distribution of their proportional consumption. | ||
|
||
- **CPU** - Consider the share of vCPUs reserved compared to the total number of physical cores available. | ||
- **RAM** - Consider the share of allocated RAM compared to the total RAM of the hypervisor. | ||
- **Storage** - Consider the use of allocated storage compared to the total capacity of the hypervisor. | ||
|
||
<Lightbox src="scaleway-allocation-rules-instances.webp" alt="" /> | ||
|
||
The value derived from this calculation will be used in every step of the calculation of the Instance's footprint. The variable is represented as `Resources_Used_VM` in the image above. | ||
|
||
Refer to the table below to understand how the `Resources_Used_VM` is integrated into the calculation of each aspect of the total footprint calculation. | ||
|
||
To obtain an accurate calculation of the Instance's footprint, we multiply the results in each step of the [basis calculation](/environmental-footprint/additional-content/environmental-footprint-calculator/), bar the Cross-IT equipment & stock calculations step, by the `Resources_Used_VM` ratio. | ||
|
||
## Cross-IT equipment & stock calculation for Instances | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. There is no "special" calculation for Instance stock |
||
|
||
The only part of the calculation that will differ from the basis is the [Cross-IT equipment & stock calculations](/environmental-footprint/additional-content/environmental-footprint-calculator/#cross-it-equipment-and-stock). | ||
|
||
To calculate the electrical consumption of Instances, instead of using an exact measure of the power consumption, we use a proxy, e.g. CPU usage. The relationship between CPU usage and the power consumption of the underlying machine is non-linear, and characterised by a consumption profile. | ||
|
||
A consumption profile is a formula that takes in a given workload (e.g. % CPU usage), and returns the estimated power consumption of the machine. | ||
|
||
We use the [consumption profiles defined by Boavizta](https://doc.api.boavizta.org/Reference/routes/#consumption-profile-routes) for the CPUs used on Scaleway Instances. | ||
|
||
Initially we base the calculation on a [theoretical value](https://doc.api.boavizta.org/Explanations/components/cpu/#usage-impacts) of 30% CPU usage. | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. for estimation data display in instance order page (cf https://www.scaleway.com/en/docs/environmental-footprint/additional-content/environmental-footprint-calculator-estimation/) |
||
|
||
### Calculation example | ||
|
||
In the example below we will calculate the manufacturing impact allocated to an Instance for a one-year usage period. | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. why not do an exemple for 100hour usage like bare metal ? |
||
|
||
Consider the following characteristics: | ||
|
||
| Resource | Instance | Hypervisor | | ||
| --- | --- | --- | | ||
| vCPU | 4 | 16 CPU cores | | ||
| RAM | 8 GB | 64 GB | | ||
| Storage | 50 GB | 1 TB (1000 GB) | | ||
| Lifespan (in years) | | 6 | | ||
| Manufacturing impact | | 100 kgCo2e | | ||
|
||
The allocation of the hypervisor resources for this particular Instance is calculated in the following manner: | ||
|
||
| Resource | Allocation | | ||
| --- | --- | | ||
| CPU | 4/16 = 0.25 | | ||
| RAM | 8/64 = 0.125 | | ||
| Storage | 50/1000 = 0.05 | | ||
| Total Instance Share | 0.25 + 0.125 + 0.05 = 0.425 | | ||
|
||
We multiply the resource allocation by the Instance's lifespan ratio and the hypervisor's manufacturing impact to obtain the manufacturing impact allocated to the Instance: | ||
|
||
``` | ||
1 / 6 x 100 x 0,425 = 7,1 kgCO2eq | ||
``` | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Can we add an exemple for the usage impact too ? for a 100hour usage period of an instance in FR-PAR-2 with a mean of 40% for CPU usage. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
May be we should precise that bare-metal = Elastic Metal and Apple silicon (and soon Dedibox)