Cloud Services has long recommended that "critical" data be backed up outside of the Cloud VPS/Toolforge environment by project admins/tool maintainers. This is a burden on our volunteers in that it requires them to make decisions about technology, location, etc as well as finding a method to fund this archival storage. Providing automatic full and incremental backups of all content with a self-service recovery interface would be ideal, but currently is outside of both the hardware and human capacity of the cloud-services-team. We can do something better than nothing in this area however.
We would like to provide a system with the following characteristics:
- quota limited, multi-tenant storage
- on-site and off-site mirroring of archived data
- self-service (or very low friction) opt-in for usage
- end-user interface via ssh and possibly rsync protocols
- rsync.net inspired flexibility for users
- How-to/tutorial documentation for basic usage
Additional storage capacity for a less ambitious backup service was added during the FY18/19 planning cycle. The cloudstore100[89] pair have disk capacity to provide the on-site storage and cloudbackup200[12] will provide space for off-site redundancy.