Cloudtik

- Seperate YARN runtime for reuse by Spark and Flink.
- Seperate Hadoop client runtime for generalizating storage.
- Seperate Mount runtime for reuse.
- Kubernetes: support permanent data volumes for stateful clustering.
- API backend discovery for Kong and APISIX.
- New MinIO runtime for local S3 compatible storage. Hadoop client and Mount runtimes to support MinIO.
- New Redis runtime with support of replication cluster.
- New MongoDB runtime with support of replication and sharding cluster.
- MySQL runtime supports replication and group replication cluster.
- Postgres runtime supports stream replication cluster.
- Docker image building reuses the runtime installation process.
- A lot of other improvements covering usability and fixes.

- Support multiple cloud object storages in one workspace for public cloud providers. And the cluster can configure to use one specific cloud storage.
- Support multiple database instances in one workspace for AWS, Azure and GCP. And the cluster can configure to use one specific cloud database instance.
- Improve cluster controller to support allocating stable sequence ID for nodes. This is useful for clusters need reusable volumes/disks.
- Support cluster with permanent data volumes for AWS, Azure and GCP utilizing stable sequence ID.