Distributed Storage Cluster Consulting & Operations
Ceph, BeeGFS and MinIO clusters designed around your real failure domains, benchmarked with fio, and proven to survive a disk, a node, and a rack going away.
What this usually looks like
Storage is the part nobody wants to own, and the part where a wrong CRUSH map costs you a weekend of rebalancing.
- Performance was never baselined, so every slowdown becomes an argument.
- Adding a disk rebalances the whole cluster during business hours.
- You have a cluster that survived a disk failure in a demo but never in production.
- Object storage exists in three places with three different S3 compatibility quirks.
Three ways to buy this
Audit if you want to know how bad it is. Build if you have the team and want the capability. Managed if you don't want to own the pager. Most engagements start with the audit and decide afterwards.
Audit
A fixed-price health check — the cheapest way in
- CRUSH map and placement-group review
- fio baseline: sequential, random, and mixed, before any change
- Failure-domain analysis: what happens when a rack loses power
- Capacity and rebalance-time projection as you grow
Build
One-off delivery: design, code, hardening, handover
- Ceph cluster (RADOS/RBD/CephFS/RGW) deployed with cephadm, tuned BlueStore and device classes
- Rook-Ceph on Kubernetes where the workloads are containerised
- Parallel filesystems (BeeGFS, Lustre) for HPC and AI workloads
- S3-compatible object tier (MinIO, Garage, or Ceph RGW) with lifecycle policy
- NVMe-oF / iSCSI / RDMA block paths for latency-sensitive databases
Managed
Build plus 7×24 on-call, SLA and quarterly drills
- Health review and rebalance scheduling outside business hours
- Capacity forecasting and OSD replacement procedure
- Performance regression detection against the stored baseline
- Upgrade path management across Ceph releases
What we actually deploy
Specific tooling, not categories. If a version matters to your decision, ask — we will tell you which one we would pick and why, including when the answer is the older one.
Unified
- Ceph (Squid / Reef)
- Cephadm + Orchestrator
- Rook-Ceph 1.17
- Ceph RGW (S3)
- CephFS
Parallel filesystems
- BeeGFS
- Lustre
- WekaFS
- JuiceFS
- GPFS / IBM Storage Scale
NAS & file services
- NFS-Ganesha
- Samba / AD integration
- GlusterFS (legacy maintenance)
Object
- MinIO
- Garage
- SeaweedFS
- Ceph RGW
- S3 lifecycle + versioning
Block & replication
- DRBD 9 + LINSTOR
- Longhorn
- OpenEBS Mayastor
- ZFS (RAIDZ, send/recv)
- LVM / mdadm RAID
Fabric
- NVMe-oF
- iSCSI
- RDMA / RoCE v2
- InfiniBand
- multipath (dm-multipath)
Tuning & validation
- CRUSH map design
- BlueStore tuning
- cache tiering / NVMe write cache
- fio
- rados bench
- ceph osd perf
What we will be measured on
These are the numbers we put in the delivery report. If we cannot measure it, we do not promise it.
- A written fio baseline report — the reference every future change is measured against.
- Documented, tested behaviour for disk loss, node loss, and whole-rack loss.
- Expansion without service interruption, with the rebalancing rate capped and scheduled.
- Data durability stated as a number, derived from your replica count and failure domains — not a slogan.
- OSD replacement as a timed, documented procedure your team can execute.
What we would build, and where the weak point is
With three racks and size=3, a whole rack can be lost. With two racks or a cluster where all MONs share one PDU, it cannot — and the dashboard will still show HEALTH_OK until it matters.
marks a single point of failure we would raise with you rather than hide. Every architecture has one; the question is whether you know where it is.
What we don't do
Every vendor says yes to everything. Here is where our answer is no, or where someone else needs to be in the room. If this list looks wrong for your situation, tell us — we would rather argue about scope before the contract than after.
- Requires partner
Storage hardware supply, warranty and replacement parts
We specify disks and chassis; your vendor supplies them. We do not resell.
- Out of scope
Recovering data from a physically failed disk
That is a clean-room forensic service. We design so you never need one, and we will tell you which backups to restore from.
- Requires partner
Tape library and long-term archive operations
We integrate to your archive tier; we do not operate tape robots.
- Out of scope
Lustre on hardware without a validated interconnect
Lustre without proper fabric behaves badly. We will recommend BeeGFS or Ceph instead rather than build something you will regret.
What it looked like last time
A Ceph cluster that had never been baselined was benchmarked with fio, found to be bottlenecked by a single mis-sized CRUSH rule, and after retuning delivered roughly 2.4× the original write throughput on identical hardware.
Anonymised under NDA. No client names, no logos, no invented metrics. Full case library on the work page.
Talk to the storage engineer about your distributed storage
Send us the architecture, or just describe what broke most recently. You'll get an engineer's answer, not a discovery questionnaire.
Talk to a storage engineer: botonwa@gmail.com · +86 130 5181 3099