.. _what_is_rcs3:

What is RCS3
============

.. warning:: THIS is UNDER DEVELOPMENT and SHOULD NOT BE USED UNTIL 'release'

:bluelight:`RCS3`:
  `Rclone to S3 for large backup`
  is developed by the `Research Cyberinfrastructure Center <https:/rcic.uci.edu>`_ (:term:`RCIC`)
  at the University of California, Irvine (:term:`UCI`).

:bluelight:`Motivation`:
   Need a simple solution to backup large quantities of data from servers
   used by different groups and labs. A simple client access to a solution
   where the setup complexity is handled by a central admin scenario.

:term:`RCS3` is a set of programs, scripts, templates, policies, and documentation for UCI's use of ``rclone``
to backup multi-Terabyte to Petabyte-scale storage servers to *Amazon S3 Glacier Deep Archive*.

In particular, :term:`UCI` plans to use these to backup large data servers in labs.
Our primary target is  the in-faculty-lab storage server that is often run by a graduate student
or other part-time administrator (sysadmin).

In the UCI environment, central IT does not have system-level privileges on the lab servers.
Central IT (cloud admin) provisions per-server backup buckets, applies appropriate permissions and policies,
and provides initial guidance on setting up daily backups.

The lab sysadmin configures their system for backup using buckets and credentials created by the cloud
admins. Lab sysadmins can initiate restores to that backup bucket, but do not have the authority to permanently
delete backups. Intentional administrative isolation means that complete data loss requires two
different sets of elevated privilege credentials -
one to delete primary data on the server, and a different set to permanently delete data on the backup.

Object versioning in S3
with specific retention policies provide a delay (e.g., 90 days) from delete/overwrite to permanent deletion. This
essentially provides "snapshots" so that restores can be accomplished from any recorded state during the
retention period.

:term:`RCS3` is command-line driven and demands a small number of common
software prerequisites (see :ref:`requirements` for details): 
Rclone, Python3, selected Python3 packages (e.g., boto3 tools, PyYaml, multiprocessing) installed on the to-be-backed-up server. 

We've successfully used these scripts and policies on Linux, Windows, and Intel-based Synology NAS.

We're providing this information AS IS with no guarantees. Our hope is that others (especially other academic
institutions) can use our model as a starting point to create cost-effective and robust backups for their research
communities.

This software is released under the *GNU General Public License v3.0*.