Rook Ceph Storage for Kubernetes Clusters
Redundant Block, File, and Object Storage
Kubernetes-Native Alternative to NAS and SAN
The Rook storage orchestrator is a graduated CNCF project since 2021. The Rook operator facilitates the creation and management of Ceph clusters within your Kubernetes cluster. It provides CSI drivers for the creation of RBD (block), CephFS (file), and S3 / Swift (object) storage through persistent volume claims (PVCs) and object bucket claims (OBCs) using your raw HDD or SSD storage.
Compared to NAS and SAN offerings from vendors such as Dell EMC, NetApp, and Synology, Ceph is a 100% open source solution. There are no costly maintenance contracts, and the storage can scale to exabytes without a per-TB cost. Ceph has been adopted by customers including CERN, Comcast, Ford Motor Company, and SpaceX.
Worried about GlusterFS EOL? Switch to Ceph.
Kubernetes storage solutions including GlusterFS, Longhorn, and StorageOS are reaching end-of-life (EOL). These projects are no longer receiving updates or only maintenance (security) updates.
Ceph has become the de-facto standard for Kubernetes storage due to its incredible scalability and support from industry heavyweights including IBM and Red Hat.
Also, Rook Ceph is cloud-agnostic meaning that you can deploy it using block volumes from any cloud provider, or storage directly on the host in a bare metal or OpenStack environment.
ReadWriteMany (RWX) Persistent Volumes for Kubernetes – better performance than NFS
Migrate virtually any application to Kubernetes, even non-cloud native applications, using ReadWriteMany (RWX) persistent volumes that can be attached to multiple pods on different nodes – without the risk of data corruption or locking issues.
CephFS persistent volumes support much higher throughput (hundreds of MB/s) than legacy NFS or CIFS based solutions such as Amazon EFS or Azure Files. Also, there is no 1 TB/mo minimum unlike with Google Cloud Filestore.
For best results, a Ceph cluster should have a minimum of 3 nodes with a 1Gbe LAN (10 Gbe preferred) and SSD storage (NVMe preferred). Our Ceph consultants can advise you on the specifications for your use case.
Data Safety Through Replication & Encryption
Rook Ceph is designed to protect the durability and security of your data. The most typical deployment uses a 3X replication factor where data is triple replicated to OSDs on 3 nodes before a write operation is acknowledged. For greater usable vs. raw storage, this can be reduced to a 2X replication factor with erasure coding – with the tradeoff of higher RAM & CPU usage.
To ensure security & compliance, both at-rest encryption and over-the-wire encryption are available with Ceph. Industry-standard LUKS and dm-crypt based encryption is supported for both entire OSDs (raw devices) and using custom passphrases for individual RBDs (block devices).
