Ceph has released version 20.2.0, code-named “Tentacle”, marking the 20th stable release of the platform. This update brings meaningful improvements to performance, efficiency and day-to-day operations, especially for teams running Ceph in production at scale.

For operators, architects and engineers working with block, file or object workloads, Tentacle is a practical release with clear benefits: faster erasure coded pools, better RBD workflows, improved CephFS behavior and more predictable management tooling.

Why this release is worth your attention

Ceph Tentacle focuses on improvements that matter in real clusters:

• higher performance in erasure coded pools
• safer and more predictable operational behaviour
• smoother imports and migrations for block devices
• better compatibility with S3 workloads
• cleaner identity and management integration
• more structured multi cluster operations

Tentacle is not a cosmetic update. It changes how clusters behave under load and how teams manage daily operations.

Key improvements in Tentacle

RADOS

• FastEC introduces substantial optimizations for erasure coded pools, improving partial reads and writes and reducing space amplification.
• ISA-L becomes the default EC pluginfor new pools, replacing the unmaintained Jerasure.
• BlueStore receives compression and WAL performance improvements.
• A new tech preview command, ceph osd pool availability-status, provides direct insight into pool availability.

RBD

• RBD images can now be imported instantly from another cluster or external sources via NBD stream.
• Namespace remapping adds flexibility and control for RBD mirroring.
• The Python API gains timezone aware timestamps.

CephFS

• Directory behavior can be configured to be case insensitive or normalized, helpful for mixed-OS environments.
• Safer operations when modifying max_mds on unhealthy clusters.
• Improved subvolume info and safer namespace pool naming conventions.

Object Gateway (RGW)

• Adds support for S3 GetObjectAttributes.
• More efficient bucket resharding, with heavy processing moved before the write-blocking stage.
• Switch from tenant based IAM to proper account semantics for cleaner identity management.

cephadm, dashboard and management

• New mgmt-gateway service provides a centralized TLS termination point for Dashboard, Prometheus, Grafana and Alertmanager.
• OAuth2/OIDC support in Dashboard and monitoring tools.
• Support for NVMe/TCP gateway groups, multiple namespaces and improved SMB for CephFS.
• Removal of old, deprecated modules like mgr/restful and mgr/zabbix.

Upgrade notes and best practices

Before upgrading:

• Check your upgrade policy (42on advises latest versions –1).
• Verify your cluster is fully healthy
• cephadm users can upgrade with:  ceph orch upgrade start –image quay.io/ceph/ceph:v20.2.0
• non cephadm deployments should upgrade services in sequence: Mon, Mgr, OSD, MDS, RGW
• downgrades from Tentacle are not supported

Operational recommendations from our work at 42on:

• Test the upgrade in a staging environment with real-world data patterns.
• Benchmark EC performance before and after the upgrade, FastEC can bring meaningful gains.
• Update operational runbooks to include the new pool availability command.
• Communicate changes like the switch to ISA-L with your internal teams.

Where Tentacle fits in real workloads

For engineering teams managing large, mixed workloads, analytics pipelines, virtual machine storage, archiving, large buckets or simulation data, Tentacle’s improvements are tangible. Faster EC pools reduce cost-per-TB tradeoffs, improved RGW behaviorhelps with large object sets and the new identity and management tooling simplifies integration with existing infrastructure.

A brief note on open source and long-term direction

Tentacle continues Ceph’s growth as a mature open source platform with a steady release cadence. While not the focus of this article, these improvements also support organisations that value control and long-term independence, but the practical operational gains remain the real highlight of this release.

Final thoughts

Ceph v20.2.0 is a practical, operator-friendly release with benefits across performance, cluster stability and management. If you run Ceph today, this is a release worth planning for. If you’re evaluating Ceph, Tentacle is a strong point in its maturity curve.