The greatest advantage with a tmpfs is that you can use the faster access Volatile memory to store files instead of Secondary storage system.
Lets head off to the Demonstration using tmpfs.
Create a Directory named test-docs in /mnt
[vamshi@SERVER02 mnt]$ sudo mkdir /mnt/test-docs
We will be using the dd command to dump the data and create a file with given Block size, as below and compare the write speeds on tmpfs (Linux temporary Filesystem) vs conventional storage filesystem.
[vamshi@SERVER02 mnt]$ sudo mount -t tmpfs -o size=1G tmpfs /mnt/test-docs
Output from the mount command:
[vamshi@SERVER02 mnt]$ df -hT /mnt/test-docs/ Filesystem Type Size Used Avail Use% Mounted on tmpfs tmpfs 1.0G 0 1.0G 0% /mnt/test-docs Now add the entry to /etc/fstab to make the mount persistent across reboots(We should write How to create an fstab entry in another article)
tmpfs /mnt/webdocs tmpfs nodev,nosuid,noexec,nodiratime,size=1G 0 0
With these changes it is going to perform faster system read writes during the runtime.
Please advise caution as this uses the RAM.
Now to test the setting Temporary Filesystem vs Storage Filesystem write speeds
We will demonstrate see about the temporary(tmpfs) filesystem on linux in action and practical examples.
Writing a 500MB file on the same server to a Secondary storage vs Writing it to tmpfs.
Lets us write a single 500MB file and measure its write speed and time it’s operation
[vamshi@SERVER02 mnt]$ time sudo dd if=/dev/zero of=/mnt/test-docs/dump.txt bs=1k count=500000 500000+0 records in 500000+0 records out 512000000 bytes (512 MB) copied, 0.815591 s, 628 MB/s real 0m0.871s user 0m0.103s sys 0m0.760s
This operation finished under 0.8 seconds time which is around 100 milliseconds lesser than 1 second and write operation completed with 628MB/s.
Lets check on the file size on the tmpfs mountpoint /mnt/test-docs/
[vamshi@SERVER02 mnt]$ df -hT /mnt/test-docs/ Filesystem Type Size Used Avail Use% Mounted on tmpfs tmpfs 1.0G 489M 536M 48% /mnt/test-docs
Now lets run the same write operation to the secondary storage filesystem
[vamshi@SERVER02 mnt]$ time sudo dd if=/dev/zero of=/mnt/dump.txt bs=1k count=500000 500000+0 records in 500000+0 records out 512000000 bytes (512 MB) copied, 1.38793 s, 369 MB/s real 0m1.510s user 0m0.028s sys 0m1.227s
The write operation to the disk took twice the time at a write speed of 369MB/s
The space usage on the [/code]/mnt/test-docs/[/code] after the operation is 489MB out of 1024MB.
[vamshi@SERVER02 mnt]$ df -hT /mnt/test-docs/ Filesystem Type Size Used Avail Use% Mounted on tmpfs tmpfs 1.0G 489M 536M 48% /mnt/test-docs
CONCLUSION:
The write/read speeds to tmpfs were substantially faster compared to the secondary storage block device mount points, The fact of the matter is tmpfs although the clear winner is tmpfs, it cannot be used for persistent data storage as its just sits on top of the memory and in no way offers long term data storage persistence.
So what is a practical usage of tmpfs in linux ?
The practical usage can be storing of the static web content to serve the images,css and js to fasten the request speeds.
The secondary storage as we know for straight persistence and long term data storage capabilities.