Ceph PG Calculator — Placement Group Sizing for Your OSD Topology

Free Ceph placement group calculator for storage administrators and OpenStack engineers. Enter your OSD count, pool count, replication factor, and target PGs per OSD to get mathematically correct pg_num and pgp_num values for every pool in your Ceph cluster — with the exact ceph CLI commands to create them. No sign-up required.

What this calculator computes

Correct pg_num and pgp_num per pool snapped to the next power of 2
Per-pool PG distribution weighted by pool weight inputs
Total PGs across all pools with actual PGs per OSD ratio
Raw cluster capacity vs usable capacity accounting for replication and EC overhead
Storage utilization percentage with warning at 80 percent threshold
Erasure code pool overhead — k+m divided by k storage efficiency factor
CRUSH failure domain awareness for host, rack, and datacenter topologies
Warnings for under-distributed PGs, memory pressure, and unsafe OSD counts

Ceph PG sizing formula

Formula: Total PGs = (OSD count × target PGs per OSD) divided by replication factor, rounded up to the next power of 2.

Per pool: Divide total PGs by number of pools, adjusted by pool weight fraction, rounded up to next power of 2 with a minimum of 32.

pgp_num: Always set equal to pg_num.

Sweet spot: 100 to 200 PGs per OSD across all pools combined. Below 50 causes uneven distribution. Above 300 causes memory pressure at approximately 10 MB RAM per OSD per 100 PGs.

Replication vs erasure code overhead

Replication size=2: 50% storage efficiency, tolerates 1 OSD failure
Replication size=3: 33% storage efficiency, tolerates 2 OSD failures (recommended)
Erasure code 4+2: 67% storage efficiency, tolerates 2 OSD failures
Erasure code 6+2: 75% storage efficiency, tolerates 2 OSD failures
Erasure code 8+3: 73% storage efficiency, tolerates 3 OSD failures
Erasure code 8+4: 67% storage efficiency, tolerates 4 OSD failures

CLI commands generated by this tool

For each pool the calculator generates the complete command set:

ceph osd pool create [name] [pg_num] [pgp_num] replicated
ceph osd pool set [name] size [replication_factor]
ceph osd pool set [name] min_size [minimum_replicas]
ceph osd pool set [name] pg_num [pg_num]
ceph osd pool set [name] pgp_num [pgp_num]
ceph osd pool application enable [name] rbd
Global config: mon_max_pg_per_osd and osd_pool_default_pg_num
Verification: ceph health detail, ceph osd df tree, ceph pg stat

Common Ceph PG sizing mistakes

Not rounding pg_num up to the next power of 2
Setting pgp_num different from pg_num
Too few PGs per OSD causing uneven data distribution during recovery
Too many total PGs consuming excessive OSD memory
OSD count less than 3 times the replication factor
Not accounting for erasure code storage overhead in capacity planning
Exceeding mon_max_pg_per_osd limit causing HEALTH_WARN

Ceph documentation

Enable JavaScript to use the interactive Ceph PG calculator and generate your exact pool creation commands.

Ceph PG Calculator

Stop guessing at placement group counts. Enter your OSD topology and get mathematically correct pg_num and pgp_num values for every pool — with the exact CLI commands to create them.

Nova / OpenStack

Replicated + EC

CRUSH Aware

Multi-pool

Capacity Planning

Free · No Login

Cluster Topology

OSD Count // total in cluster

Planned Pools // total

Replication / Protection // default for pools

size=2

size=3

size=4

Erasure Code

Target PGs per OSD // Ceph recommends 100–200

Total Data (TB) // expected

OSD Capacity (TB) // each

CRUSH Failure Domain

host

rack

datacenter

osd

Pool Names & Weights // name · weight · type

PG Sizing Rules

Formula:(OSDs × target_pgs_osd) / rep_factor

Round:Always round UP to next power of 2

Per pool:Total PGs × pool_weight_fraction

Minimum:8 PGs min; 32 recommended per pool

pgp_num:Always set equal to pg_num

Sweet spot:100–200 PGs per OSD total

EC overhead:(k+m)/k × raw storage consumed

Memory:~10 MB RAM per OSD per 100 PGs

Documentation

Placement Groups — Official Guide ↗
Complete Ceph PG administration documentation
Managing Pools ↗
ceph osd pool create, set, ls detail
Erasure Code Profiles ↗
EC plugin configurations and overhead tables
CRUSH Maps ↗
Failure domain configuration and tuning

Calculation Results

configure your cluster topology on the left
and click CALCULATE PG COUNTS
to generate pool sizing and CLI commands

Total PGs across all pools

—

—

Actual PGs/OSD

—

Raw Capacity

—

Usable Capacity

—

Storage Util.

—

Replication

—

EC Efficiency

—

Pool name	Type	Weight	pg_num	pgp_num	PGs/OSD

PG Distribution Heatmap

PGs / OSD

—

Storage utilization

—

CRUSH domain

—