Thin Provisioning (TP) operates by allocating disk storage space in a flexible manner among multiple devices, based on the minimum space required by each volume at any given time.
Test Case
- On the Lightbits server, check the total free physical capacity of storage.
- Create volumes, in which the total logical size is bigger than the free physical size of storage.
- Prove that each client can allocate volumes that exceed the total physical capacity of the resource pool.
- Connect these volumes on the client side (reflected as a new namespace on the client side), and use FIO to test that they work properly.
- On the Lightbits server, check the freePhysicalStorage capacity parameter, to find out how much free storage capacity is available:
$ lbcli get cluster -o json
In this example, we have 27.8 terabytes physically available, which means that you can create a total of X terabytes of logical volumes.
- Run a simple ‘for’ loop in order to create a few volumes, whose total capacity are greater than the ‘freePhysicalStorage’ parameter. In this example, we will create six volumes of 12 Tib each; 6x12=72 Terabytes.
Sample Output
- From the client server, connect to the newly-created volumes and check that six new NVME devices of 12 Tib each are attached, with
lsblk:
- Run random FIO tests on the volumes and check performance:
- Make sure you have FIO installed. If not:
$ yum install fio -y - With lshw, double-check the num of CPUs and apply accordingly, on
numjobs=32 cpus_allowed=0-31in the FIO file.
Sample FIO file:
Was this page helpful?