The stack provides many possibilities. This page attempts to expose them.
Modularity and flexibility
Half generic, half specializedBlueBanquise is split between 2 roles collections:
- Core: the CORE of the stack, focused on providing roles and tools to deploy hosts and configure vital services. CORE is generic to any kind of cluster.
- Community: COMMUNITY roles and tools provides additional production level features on top of CORE to specialize cluster.
Ansible roles collections around a data-modelBlueBanquise can be seen as a set of Ansible roles collections, which ensure full modularity.
Even elements of the core can be replaced by others.
A data model ensure coherency between roles.
The Ansible inventories structure used can handle a wide range of cluster and network architectures, and allows adding more variables as desired.
Compatible and modernThe whole stack is based on python 3.6+ and Ansible, for better long term investment.
The PXE stack is Legacy or EFI compatible, and should allow to boot any kind of hardware.
A focus is made on reducing scripting to the minimum, allowing stack to grow and extend easily.
All stack data are stored in plain text files, in YAML format, using Ansible inventory folders structure.
No binaries + no databases = fully editable and tunable with minimal knowledge.
Supported OS and platformIn its current state, stack can deploy on mixed Linux distributions and platform (in the same cluster), including:
- RHEL and derivated, versions 7.X and 8.X
- Ubuntu version 20.04
Stack is compatible with x86_64 and arm64/aarch64 platforms.
This ability to handle multiple Linux distributions and architectures at the same time, allows to easily handle exotic hardware or cluster needs.
RolesThe stack provides many Ansible roles, seen as independent elements. Roles are autonomous from each other's, to be easily switched with externals or user's custom roles.
Specific advanced featuresThe stack also as few native mechanisms under the hood. These features are a need for us, so we added them.
Multi IcebergsThis feature is sometime called multi islands in HPC or Cloud.
BlueBanquise can split cluster into independent and autonomous groups of equipment. These new subclusters are called icebergs, each iceberg being managed by a group of management nodes and isolated from the others (but can be reached through interconnect if exist and asked).
This feature is mainly used:
- When there is a need to separated parts of the cluster.
- When there is a need to run an old cluster aside a new one keeping both in production.
- When there is a need to scale to very large number of hosts.