| layout | title | permanlink |
|---|---|---|
page |
About |
/ABOUT/ |
We are a collective group of Ceph and storage enthusiasts who are looking to find ways to simplify the Ceph installer process. We are redefining what it means to create a storage appliance, while keeping to the basics of building something that can provide and manage data without an application context.
Currently, we see Ceph as being too complex to use for the average user. This complexity stems from Ceph's flexibility that supports a huge matrix of use cases and the effects this has on available capacity, performance and thus availability.
All current management tooling deploys Ceph bottom-up, i.e. the user must specify deployment patterns at the daemon level, create pools and crushmaps (which encode various availability and performance requirements) and only then can a user deploy their workload.
We are exploring if other user interface abstractions are practical and ultimately an improvement. The goal should be simplification where a user is not required to configure individual daemons or disk layouts, but can provide a high level specification of what they want (in terms of availability, usable capacity, ...?) and a piece of code translates that to a deployment layout. Whether this deployment layout can be set up on the current hardware can be determined by software and the user can be provided with the feedback (be it positive or negative).