Gen3
The content covers Gen3 to the new bwCloud-OS infrastructure. For the legacy version (Gen2), see Flavors (Gen2).
Available Flavors
A flavor describes a predefined profile consisting of computing power, memory, and storage (CPU, RAM, disk), thereby defining the basic resources of a virtual machine.
Flavors enable the efficient use of available resources according to the principle of minimal resource usage and allow for a tiered selection of system configurations—ranging from small test environments to high-performance production systems.
In bwCloud-OS, instances (VMs) can be upgraded to a larger flavor as needed, provided that the assigned quota flavor (usage limit) allows it. In addition, the storage space of an existing VM can be expanded by attaching additional volumes (block storage). This allows computing power and storage capacity to be flexibly adapted to current needs without having to create a new VM.
Flavors define the basic configuration of a virtual machine in terms of vCPUs, RAM, and storage. This directly determines the consumption of booking units (BEH), which serves as the basis for cost allocation in bwCloud-OS. The conversion of BEHs into euro amounts takes place exclusively between bwCloud-OS and the participating partner institutions and is therefore not published here.
Quota-Flavor
The Cost Cap for Orders
Quota-Flavor = Project-Quota-Flavor
A quota flavor defines a project’s maximum resources - that is, the total amount of resources that all VMs in a project are allowed to use collectively (e.g. total number of vCPUs, RAM, and storage).
The quota flavor should be assigned to a user by their home institution before registration.
Based on this, the initial project quota is set automatically.
Within the bwCloud-OS Dashboard, VMs can only be created within the limits of the quota flavor.
Flavor
Resource Reservation / Allocation
Flavor = VM-Flavor (as shown in the table above)
A flavor defines the resources of a single virtual machine (vCPUs, vRAM, vDisk).
It is selected when creating a VM and determines its configuration and resources.
In general, additional resources can also be assigned to a VM at a later time.
However, this requires sufficient available project quota for the flavor to be assigned.
Sample Use Cases for Flavors
The following examples can help you choose among the different flavors and make it easier to decide which one is best suited for various use cases.
This overview provides guidance for small programming exercises, web servers, databases, and resource-intensive applications:
Linux Basics and Programming
(e.g. Teaching and Training)
Recommendation:
Flavors with 1 vCPU and up to 2 GB of RAM.
These resources are sufficient for learning the basics of Linux or running simple programs and scripts. You can also set up small test servers or temporary development environments to try out new software or configurations.
The low CPU and memory requirements make these flavors particularly cost-effective for learning and testing purposes.
Web servers
(e.g. small educational or project websites, or test systems)
Recommendation:
Flavors with 2–4 vCPUs and 4–8 GB of RAM.
These configurations are well-suited for web servers such as Apache or Nginx, even with moderate user traffic and database connections.
The standard 20 GB system disk is sufficient for small to medium-sized web projects.
Database servers
(e.g. development databases, SQL)
Recommendation:
Flavors with at least 4 vCPUs and 8 GB of RAM.
Databases (e.g., MySQL, PostgreSQL) benefit from more RAM (for larger data volumes) and CPU cores (for many concurrent queries).
The standard system disk (20 GB) is usually sufficient for development purposes; for larger amounts of data, additional volumes (block storage) can be mounted.
Resource-intensive applications
(e.g. AI model validation, scientific computations)
Recommendation:
Flavors with 16 vCPUs and 32 GB of RAM or (upon request) more.
Such applications require high computing and memory resources for parallel processing or large datasets.
The system disk (20 GB) can be expanded with additional volumes to efficiently store data or models.