Connect, collaborate, and share expertise on IBM Storage Ceph
In modern enterprise environments, file storage is rarely used by a single workload or team. SMB deployments often serve a mix of users, applications, analytics jobs, and backup workloads each with different performance expectations. Ensuring predictable performance in such shared environments is a constant challenge.
To address this, IBM Storage Ceph 9.0 introduces SMB Rate Limiting (Quality of Service – QoS). This feature enables administrators to control SMB bandwidth usage, prevent noisy-neighbor scenarios, and deliver consistent performance across workloads.
In this blog, we’ll walk through the motivation behind SMB QoS, the problems it solves, its high-level architecture, and how it works in Ceph 9.0.
Without QoS, SMB workloads operate on a first-come, first-served basis. This can lead to several issues:
A single heavy workload, such as a backup or large file copy, can consume most of the available bandwidth
Other users may experience degraded performance or timeouts
Performance becomes unpredictable and difficult to troubleshoot
Meeting SLAs in multi-tenant environments becomes challenging
As SMB adoption grows in enterprise and cloud deployments, these limitations become more visible and impactful.
SMB Rate Limiting, or QoS, allows administrators to control the bandwidth consumed by SMB shares. By enforcing limits at the share level, Ceph ensures that no single workload can overwhelm the system.
With SMB QoS, administrators gain:
Better control over SMB traffic
Predictable and consistent performance
Protection for critical workloads
Improved overall cluster stability
This capability is natively integrated into IBM Storage Ceph 9.0, eliminating the need for external traffic-shaping tools or complex per-node tuning.
The primary goals behind introducing SMB Rate Limiting in Ceph 9.0 are:
Workload isolation – Prevent one SMB workload from impacting others
Predictable performance – Ensure consistent throughput and response times
Enterprise readiness – Support multi-tenant and mixed workload environments
Simplified management – Centralized configuration and enforcement
These goals align with real-world enterprise use cases where SMB is used for diverse and performance-sensitive workloads.
The SMB QoS workflow in Ceph 9.0 is straightforward:
A storage administrator creates an SMB cluster and one or more SMB shares
QoS values are configured per share using:
Imperative CLI commands, or
Declarative spec-based configuration
The updated QoS configuration is applied dynamically
SMB client traffic to that share is rate-limited based on the configured values
Currently, QoS configuration is supported per SMB share, providing fine-grained control over bandwidth usage
Updating QoS value
Supported QoS parameter
Imperative style cli command:
ceph smb share update <cephfs volume> qos <smb cluster id> <share id>\ --read-iops-limit=100 \ --write-iops-limit=200 \ --read-bw-limit=1048576 \ --write-bw-limit=2097152 \ --read-delay-max=5 \ --write-delay-max=5
Example:
# ceph smb share update cephfs qos smb1 share1 --read_iops_limit 100 --write_iops_limit 100 --read_bw_limit 1048576 --write_bw_limit 2097152
Declarative Style Spec File
- resource_type: ceph.smb.cluster
cluster_id: smb1
auth_mode: user
user_group_settings:
- source_type: resource
ref: ug1
placement:
label: smb
- resource_type: ceph.smb.usersgroups
users_groups_id: ug1
values:
users:
- name: user1
password: passwd
groups: []
- resource_type: ceph.smb.share
share_id: share1
cephfs:
volume: cephfs
subvolumegroup: smb
subvolume: sv1
path: /
qos:
read_iops_limit: 300
write_iops_limit: 300
read_bw_limit: 1048576
write_bw_limit: 2097152
read_delay_max: 30
write_delay_max: 30
# ceph smb apply -i /tmp/tmpg1n4547q.yaml
Disable QoS value
Imperative sytle cli method
ceph smb share update cephfs qos <share_name> <share_id> \ --read-iops-limit=0 \ --write-iops-limit=0 \ --read-bw-limit=0 \ --write-bw-limit=0 \ --read-delay-max=0 \ --write-delay-max=0
# ceph smb share update cephfs qos foo bar \ --read-iops-limit=0 \ --write-iops-limit=0 \ --read-bw-limit=0 \ --write-bw-limit=0 \ --read-delay-max=0 \ --write-delay-max=0
Declarative way spec file:
read_iops_limit: 0
write_iops_limit: 0
read_bw_limit: 0
write_bw_limit: 0
read_delay_max: 0
write_delay_max: 0
SMB Rate Limiting in IBM Storage Ceph 9.0 delivers several key benefits:
Predictable SMB performance
Effective workload isolation
Enterprise-grade bandwidth control
Simplified and centralized management
By integrating QoS directly into the SMB service, Ceph 9.0 makes SMB deployments more robust, scalable, and production-ready.
As enterprise environments continue to grow and diversify, controlling SMB workload behavior becomes essential. SMB Rate Limiting (QoS) in IBM Storage Ceph 9.0 addresses this need by providing native, share-level bandwidth control that is easy to configure and manage.
With this feature, organizations can confidently deploy SMB in multi-tenant and mixed workload environments while maintaining predictable performance and operational simplicity.
Copy
By
Mohit Bisht
posted