Replicated SDS with DRBD & LINSTOR
Presentation of the LINSTOR SDS system, which automates the management of different types of block storage volumes on a cluster of storage servers, with a focus on replicated storage for high availability and disaster recovery.
Focus on architecture & implementation of the system, integration with other software and the development process.
Intended audience:
- Operators, Architects, Developers of cloud & virtualization environments
- Operators of systems based on software defined storage
- Operators of OpenStack, OpenNebula, Proxmox virtualization platforms
- Operators of Kubernetes/Docker containerization
Focus of the lecture & discussion:
- Existing storage technologies, focus on Linux: block devices in general, device mapper/LVM, ZFS, DRBD, dmcrypt, etc.
- Challenges of automating storage management (allocation/modification (resizing, ...)/deallocation of volumes)
- Challenges of managing clusters (parallelization, networking, ...)
- Our solutions - how the system works, protocols involved, security model, configuration persistence, transaction safety, etc.
- Some interesting implementation details
- How the development team works - methods, processes, project management, the tools that we like to use
Speakers:
Robert Altnoeder