All systems operational · 99.99% uptime SLA
Console Support Docs Blog Contact →
File Storage — Shared Filesystem · NFS · SMB · POSIX Compatible

Mount once.
Access from
everywhere.

NubexCloud File Storage is a distributed shared filesystem you mount on any host via NFS or SMB — no API calls, no code changes. Any number of servers, containers, and VMs access the same directory simultaneously, with 99.999999% data reliability and up to 100 PB per filesystem.

Create a Filesystem → Capacity vs Performance
100PBPer filesystem
NFS+ SMB protocols
∞Concurrent mounts
POSIXCompatible
Shared Filesystem /data/projects 🌐 Web Server NFS ⚙️ App Server NFS 🧠 GPU Host NFS 📦 Container NFS 🖥 Windows VM SMB 🔩 Bare Metal NFS NFS · SMB POSIX compatible mount -t nfs fs.nubexcloud.com:/data /mnt
What You Get

A shared filesystem that scales with your infrastructure

File Storage gives your hosts, containers, and VMs a shared directory they all read and write simultaneously — with no data synchronisation, no API integration, and no application changes.

8 nines
Data reliability
Three-replica mechanism with cross-cabinet deployment. Immune to single-disk and single-machine failures. Controllable MTTR with priority reconstruction.
100 PB
Per filesystem
Single filesystem scales to 100 PB. Online expansion is transparent to applications — mount point stays the same, capacity increases in the background.
∞
Concurrent mounts
No limit on the number of hosts that can mount a filesystem simultaneously. VMs, containers, physical servers, and managed cloud hosts all supported within the same VPC.
POSIX
Full compatibility
NFS and SMB protocols with complete POSIX semantics. Existing applications connect without any code changes — standard file operations work exactly as expected.
How It Works

Mount. Use. Share. Three commands.

File Storage behaves exactly like a local disk — because to your applications, it is one. Mount it once and every tool, language, and framework that works with files works immediately.

1 Create filesystem in console
✓ Name: my-shared-fs
✓ Type: Performance · SSD
✓ Region: Dubai
✓ VPC: vpc-prod-dubai
→ Mount point: 10.0.0.88:/data
2 Mount on any host (Linux)
$ mkdir -p /mnt/shared
$ mount -t nfs \
  10.0.0.88:/data /mnt/shared
$ df -h /mnt/shared
Filesystem   Size  Used Avail
10.0.0.88:/data 10T  2.4T  7.6T
3 Use it like any local directory
$ ls /mnt/shared/
models/  datasets/  logs/  code/

$ python3 train.py \
  --data /mnt/shared/datasets/
Training started · reading from shared FS
The same path · visible on all servers simultaneously
Server A $ ls /mnt/shared models/ datasets/ logs/ Server A $ cat logs/app.log [2024] Starting training... Epoch 1/100: loss=2.41 Server B $ ls /mnt/shared models/ datasets/ logs/ Server B $ cat models/v2.pt Loading checkpoint... Accuracy: 94.2% Same filesystem 10.0.0.88:/data NFS · VPC network · encrypted Write on Server A → visible on Server B instantly
Why no code changes are needed
✓
Full POSIX semantics — open(), read(), write(), seek(), stat() all work as expected
✓
Standard directory structure — same paths, same permissions model, same file tools
✓
Windows SMB access — map as a network drive, accessible from Windows Explorer and tools
Features & Types

Four core capabilities. Two performance tiers.

File Storage combines reliability, elasticity, sharing, and ease of use in every instance — then offers two distinct performance profiles to match your workload's requirements.

Three-Replica Reliability

Every file is stored across three replicas distributed over separate physical cabinets. A single disk or server failure cannot cause data loss. The system reconstructs to full redundancy automatically with controllable MTTR — 99.999999% data persistence guaranteed.

Transparent Elastic Scaling

Expand filesystem capacity online at any time — your applications, mounts, and file paths are completely unaffected. Scale from initial capacity to hundreds of petabytes without remounting or downtime. Business is unaware of the expansion.

Unlimited Shared Access

Any number of hosts within the same VPC can mount the same filesystem simultaneously — cloud hosts, physical servers, containers, and managed cloud instances. No per-node licensing, no connection limits. All see the same data in real time.

Zero Integration Effort

NFS and SMB protocols with full POSIX semantics. Connect without changing a single line of code. Use any existing file management tool, monitoring system, or backup tool directly. Works from Linux, Windows, and containers out of the box.

Throughput · scales with capacity 1 TB 120 Mbps 5 TB 200 Mbps 10 TB 300 Mbps Max 20 Gbps Throughput = 20 Mbps/TB × capacity + 120 Mbps
Capacity Type · SATA Media
High Throughput
For sequential, large-file workloads
Media SATA HDD
Max throughput 20 Gbps
Base throughput 120 Mbps
Best for: Log backup, database backup, Docker image registry, web files, data archiving, enterprise documents
IOPS & Latency benchmarks · 4K random 4K Write 17K IOPS · 8ms 4K Read 23K IOPS · 5ms Mixed 50% 10K IOPS · 7ms Max throughput 10 Gbps Queue depth: 128 · SSD media · sub-10ms guaranteed
Performance Type · SSD Media
Low Latency
For random I/O, AI, and real-time workloads
Media NVMe SSD
Max throughput 10 Gbps
Read IOPS / latency 23K / 5ms
Best for: AI data analysis, big data processing, container storage, code hosting, real-time applications
Architecture

One filesystem. Every host sees the same files.

File Storage sits inside your VPC and acts as a central shared filesystem. Any host — cloud VM, physical server, container, or managed cloud instance — mounts the same endpoint and sees the same directory in real time.

VPC · vpc-prod-dubai · 10.0.0.0/16 Cloud VM · Web /mnt/shared Cloud VM · App /mnt/shared GPU Host · AI /mnt/datasets Container · k8s /mnt/shared Bare Metal /mnt/shared Windows VM · SMB NFS · TCP SMB File Storage Distributed Filesystem 10.0.0.88:/data /data ├─ models/ ├─ datasets/ └─ logs/ NFS + SMB · POSIX 99.999999% data reliability Cluster A · Zone 1 SATA / SSD replica Cluster B · Zone 2 SATA / SSD replica Cluster C · Zone 3 SATA / SSD replica Management Console Create / delete filesystem Manage mount points Monitor throughput / IOPS Online expand capacity Recycle bin management Single filesystem capacity 100 PB online expansion · app unaware
Use Cases

Match your workload to the right storage type

Capacity Type is optimised for high throughput with sequential access. Performance Type is optimised for low latency and random I/O. Both share the same filesystem features — choose by workload pattern.

Capacity Type · SATA · High Throughput
Sequential & Large-file Workloads
Up to 20 Gbps · scales with capacity
📋
Log Backup and Aggregation
Multiple app servers write logs to a shared filesystem. A single log aggregator reads from one mount point without per-server access setup.
🗄️
Database Backup
Database backup jobs write large sequential files to File Storage. High sequential throughput makes full and incremental backups fast without impacting production performance.
🐳
Docker Image Registry
Store container images on a shared filesystem. All nodes in a Kubernetes cluster pull from the same image storage — no per-node image duplication.
🏢
Enterprise Data Management
Internal portals, shared document repositories, and data archiving where large numbers of files are stored and periodically accessed. Replaces NAS appliances with no reconfiguration.
Performance Type · SSD · Low Latency
Random I/O and Real-time Workloads
23K IOPS · sub-5ms reads · 10 Gbps max
🧠
AI Data Analysis
Training jobs read thousands of small samples randomly from a shared dataset. Multiple GPU nodes access the same data store simultaneously without coordination overhead.
📊
Big Data Processing
Spark or Flink jobs running across a cluster all access the same input data and write to the same output location. No data copying between nodes.
📦
Container-Oriented Data Sharing
Stateful containers and microservices share persistent data through a mounted filesystem. Pods rescheduled to different nodes automatically reconnect to the same data.
💻
Code Hosting and CI/CD
Source code repositories and build artefacts stored on a shared filesystem. CI runners mount the same repository state simultaneously without checkout duplication.
Global Network

A truly global infrastructure for fast, reliable service delivery.

26
Regions
33
Availability Zones
25ms
Regional latency
99.95%
SLA uptime
Active region
Hub region (Dubai HQ)
Backbone link
Customer Stories

Teams running shared filesystems on NubexCloud

Performance Type · AI Research
8×
faster GPU training data loading
A 48-GPU training cluster previously copied datasets to local disks on each node before starting. After mounting File Storage directly, all nodes read from the shared mount simultaneously — eliminating hours of data preparation per experiment.
Head of AI Infrastructure
AI research lab, UAE · 48-GPU cluster
Capacity Type · Container Platform
0
data sync issues across 200 pods
Running 200 container pods across a Kubernetes cluster, all reading shared configuration and writing to shared output. Previously this required Redis-based sync. With File Storage, the filesystem itself is the sync mechanism — one mount, no coordination code.
Platform Engineering Lead
SaaS platform, GCC · 200-pod cluster
Capacity Type · Enterprise Data
3 PB
migrated from NAS — zero code changes
Replaced a physical NAS device serving 150 on-premises users with NubexCloud File Storage. Mounted the same way, same paths, same permissions. Applications were updated with a config entry change — no code was modified. Migration took one weekend.
IT Director
Financial services group, MENA · 3 PB
Trusted by teams across the region
Falcon AITradeSparkMasaarNEXAGENSalam DigitalOrbita
FAQ

Common questions about File Storage

Create your first shared filesystem today

Mount once. Access from everywhere.

8-nines reliability. Up to 100 PB per filesystem. NFS and SMB. POSIX compatible. Unlimited concurrent mounts. Two performance tiers — from sequential backup throughput to SSD low-latency IOPS. All inside your VPC.

Create a Filesystem → Read the Docs