Skip to content

Add a new server archetype

This guide will help you add a new server to the server archetypes. To learn more about server archetypes, please refer to the archetypes documentation.

Server archetypes CSV file

All available servers are stored in a CSV file named servers.csv located at boaviztapi/data/archetypes/servers.csv.

Column name Required Unit Description Example
id Required Server identifier platform_compute_medium
manufacturer Server manufacturer
case_type Required Server case type rack
CPU.units Required unit Number of CPU 2
CPU.name 1 CPU name
CPU.core_units unit Number of CPU cores per CPU 24
CPU.die_size_per_core mm² Die size per CPU core 8
CPU.die_size mm² Die size per CPU 500
CPU.tdp Watt Thermal design power (in watt) 150
CPU.vcpu unit Number of vCPU per CPU 32
RAM.units Required unit Number of RAM 1
RAM.capacity GB RAM quantity 1000
SSD.units Required unit Number of SSD 0
SSD.capacity GB SSD storage quantity 0
HDD.units Required unit Number of HDD 0
HDD.capacity GB HDD storage quantity 0
GPU.units Required unit GPU quantity 3 0
GPU.name GPU name
GPU.vram GB GPU memory capacity 2
POWER_SUPPLY.units Required unit Number of power supply4 2
POWER_SUPPLY.unit_weight Required kg Power supply weight4 2.99;1;5
USAGE.time_workload Required % Time workload 5 50;0;100
USAGE.use_time_ratio Required /1 Use time ratio6 1
USAGE.hours_life_time Required hours Hours life time 35040
USAGE.other_consumption_ratio Required /1 Other consumption ratio 0.33
WARNINGS Warnings

Value ranges

Some values can be inputted using ranges like the following: default;min;max. For example, if the value is 4;2;8, it means that the default value is 4 and the range is from 2 to 8.

  1. If CPU.name is set and the CPU is available in cpu_specs.csv, you do not need to fill in the other CPU attributes. The API will complete them based on the CPU.name. ↩

  2. GPU impacts are now calculated and included in the results. GPU.vram is the video RAM capacity of one card in GB. It must match the vram of the GPU in gpu_specs.csv, never an instance or server total. ↩

  3. GPU.units is the number of GPUs in the server, as the provider counts them. A card that carries two GPUs, such as the Tesla M60, counts as 2: its gpus_per_card in gpu_specs.csv converts GPUs back to cards. GPU.units must be greater than 0 for the GPU to be accounted for at all: a row with a GPU.name but GPU.units = 0 is treated as having no GPU. It is also the denominator used to allocate GPU impacts to the cloud instances hosted on this platform, so it must be at least the largest gpu_units of those instances. See Add a new cloud instance. ↩

  4. (Usually power supply duplicated so POWER_SUPPLY.units = 2. Usually POWER_SUPPLY.unit_weight is unknown, in that case use a range such as 2.99;1;5) ↩↩

  5. (Should be a range between 0 and 100 (50;0;100) without valid justification) ↩

  6. (Should be 100% without valid justification so USAGE.use_time_ratio = 1) ↩