update roadmap

Signed-off-by: Jared Watts <jbw976@gmail.com>
This commit is contained in:
Jared Watts committed 2018-03-08 19:20:39 -08:00
1 parent 66b7c93f0e
commit 18772e5315
1 file changed
+36 -26
+36 -26
View File
@@ -1,31 +1,41 @@
# Roadmap
This document defines a high level roadmap for Rook development. The dates below are subject to change but should give a general idea of what we are planning. We use the [milestone](https://github.com/rook/rook/milestones) feature in Github so look there for the most up-to-date and issue plan.
## Rook 0.7
- Ceph Block Storage (Cluster and Pool CRDs) declared beta
- Mon reliability (restarts, failing over too fast, ip changes, etc.)
- Durability of state (local storage support, config is regenerat-able)
- Run everywhere (rook-agent uses ceph-fuse, nbd-rbd / tcmu runner)
- Adding / removing nodes and disk drives (lifecycle issues, failures, etc.)
- Run with Least Privileged and possibly without privileged containers
- Shutdown / restart issues
- Improved data placement and pool configuration (CRUSH maps)
- Placement group balancer support (ceph-mgr module)
- Use new Prometheus ceph-mgr module
- Performance Testing for Block storage
- Make API and CLI Optional (plan to remove)
- Dynamic Volume Provisioning for CephFS
This document defines a high level roadmap for Rook development.
The dates below are subject to change but should give a general idea of what we are planning.
We use the [milestone](https://github.com/rook/rook/milestones) feature in Github so look there for the most up-to-date and issue plan.
## Rook 0.8
- Operator runs HA
- Ceph Object storage declared Beta
- Logging (levels, granularity, etc.)
- Support ingress controller for RGW (SSL termination)
- Support for multi-region replication for RGW
- Object storage User CRD
- Automated Upgrade
- CRD validation, cleanup, progress, status (requires engagement with API sig)
- Volume Snapshotting (consider aligning with SIG-storage)
* Multiple storage backends
* Full design
* Refactor code base and repositories to enable
* Consider support for Minio, potentially early support for other backends time permitting
* Switch CRDs to API Aggregation
* Run on arbitrary PVs
* Remove Rook API and CLI
* Migrate CI and release pipelines to a solution hosted by the CNCF
* Run with Least Privileged and possibly without privileged containers
* Shutdown / restart issues
* Support Kubernetes 1.7+ only
* Ceph features and improvements
* Disk management (1 OSD per pod, adding/removing disks)
* Placement group balancer support (ceph-mgr module)
* Mon reliability (restarts, failing over too fast, ip changes, etc.)
* Mimic support
## Rook 0.9
* Automated upgrades
* Durability of state (local storage support, config can be regenerated)
* Custom resource validation, progress, status
* Design for Volume Snapshotting and policies (consider aligning with SIG-storage)
* Ceph features and improvements
* Improved data placement and pool configuration (CRUSH maps)
* Dynamic Volume Provisioning for CephFS
* ceph-volume provisioning for OSDs
* Ceph CSI plug-in that runs everywhere (rook-agent uses ceph-fuse, nbd-rbd / tcmu runner)
* Cluster and Pool types are declared Beta
## Rook 1.0
* Declare numerous types as v1