One of the requirements for mission critical systems is to provide reliable volume backup without impacting running system. The recommended way of cinder backup is to unmount volume before backup to avoid crash consistent backup. Unmounting is intrusive in nature and may not be feasible for mission critical systems.
This talk will focus on strategy to achieve non-intrusive cinder backup. Our strategy is based on making guest OS write-through temporarily for snapshot purpose, which is practically same as unmounted volume for snapshot purpose.
We will share design and proof of concept (POC) of our work in this talk. POC is applicable to backup only. Backup produced in this manner is as per OpenStack standards so the backup is restored using OpenStack’s standard method.
The POC uses Linux Guest OS running in Nova. POC is expected to work with most of Linux Guest OSes without any change. Cinder’s snapshot feature is used internally instead of Reinventing the wheels.
Attendees will learn about backup challenges in mission critical systems. For example possibility of failure at restore when backup was taken using Cinder’s forced snapshot of attached volume. They will also learn possible ways to overcome such challenges by applying various techniques. Finally, the solution suggested by us and its advantages and disadvantages.
We will also share some of the use cases where such solution can deliver value over and above standard OpenStack backup.