Cloud Storage FUSE

Cloud Storage FUSE is an open-source adapter that allows you to mount Cloud Storage buckets as local file systems on Linux and macOS machines. It presents objects in a bucket as files and prefixes as directories, enabling applications that expect a POSIX file system interface to read from and write to Cloud Storage without modification.

---

What Is Cloud Storage FUSE?

FUSE (Filesystem in Userspace) is a Linux kernel module that allows non-privileged users to create file systems without editing kernel code. Cloud Storage FUSE (gcsfuse) uses this interface to translate file system operations (open, read, write, close) into Cloud Storage API calls.

Key Characteristics

Feature	Detail
Protocol	Translates POSIX file operations to Cloud Storage REST API calls
Platform	Linux and macOS
Authentication	Uses Application Default Credentials or service account keys
Performance	Suitable for data analytics and batch processing; not for latency-sensitive applications
Consistency	Eventually consistent for some operations; not a replacement for a POSIX file system
Cost	Free to use; you pay only for the underlying Cloud Storage operations

---

Installing Cloud Storage FUSE

On Debian/Ubuntu

# Add the gcsfuse repository
export GCSFUSE_REPO=gcsfuse-\$(lsb_release -c -s)
echo "deb https://packages.cloud.google.com/apt \$GCSFUSE_REPO main" | \
  sudo tee /etc/apt/sources.list.d/gcsfuse.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | \
  sudo apt-key add -

# Install gcsfuse
sudo apt-get update
sudo apt-get install -y gcsfuse

On RHEL/CentOS

sudo tee /etc/yum.repos.d/gcsfuse.repo > /dev/null <<EOF
[gcsfuse]
name=gcsfuse (packages.cloud.google.com)
baseurl=https://packages.cloud.google.com/yum/repos/gcsfuse-el7-x86_64
enabled=1
gpgcheck=1
repo_gpgcheck=0
gpgkey=https://packages.cloud.google.com/yum/doc/yum-key.gpg
       https://packages.cloud.google.com/yum/doc/rpm-package-key.gpg
EOF

sudo yum install -y gcsfuse

---

Mounting a Bucket

Basic Mount

# Create a mount point
mkdir -p /mnt/gcs-data

# Mount a bucket
gcsfuse my-data-bucket /mnt/gcs-data

# Verify
ls /mnt/gcs-data
df -h /mnt/gcs-data

Mount with Options

# Mount read-only
gcsfuse --read-only my-data-bucket /mnt/gcs-data

# Mount a specific "directory" (prefix)
gcsfuse --only-dir=data/analytics my-data-bucket /mnt/gcs-data

# Mount with a specific service account key
gcsfuse --key-file=/path/to/key.json my-data-bucket /mnt/gcs-data

# Mount with file caching for better read performance
gcsfuse --file-cache-capacity-mb=1024 \
  --file-cache-max-size-mb=100 \
  my-data-bucket /mnt/gcs-data

# Mount with implicit directories (create directory objects automatically)
gcsfuse --implicit-dirs my-data-bucket /mnt/gcs-data

Persistent Mount (fstab)

# Add to /etc/fstab for automatic mounting on boot
my-data-bucket /mnt/gcs-data gcsfuse rw,user,implicit_dirs,allow_other 0 0

---

Unmounting

# Unmount the bucket
fusermount -u /mnt/gcs-data

# Force unmount
fusermount -uz /mnt/gcs-data

---

Performance Considerations

Cloud Storage FUSE translates every file operation into an HTTP REST API call, which introduces latency. It is not a drop-in replacement for a local disk or NFS.

Latency

Operation	Local Disk	FUSE
Metadata (stat, ls)	Microseconds	Tens of milliseconds
Small file read	Microseconds	Tens of milliseconds
Sequential large file read	Disk-speed	Network-speed (can be fast with caching)
Random read	Fast	Slow (each seek is an HTTP request)

Optimising Performance