Identifying instance storage disks attached to VSIs in IBM Cloud is currently not straightforward on the disk driver level. The suggested solutions from IBM all have their drawbacks and are not reliable from a long-term perspective. The solutions and their respective drawbacks are as follows:
1. Use of boot ordering of devices: After deployment, boot order of devices can change based upon adding of devices and/or when instance resize is done.
2. Use Sector size of 4096: Sector size can change for disks in the long term and is not reliable.
3. IMDS metadata query: IMDS queries are too late in the boot process and instance storage disks need to be identified very early in the boot process for our use case.
4. Cache mode setting: IBM confirmed that cache mode setting can be changed in the future, so this can't be used.
Expected solution:
Adding any unique identification in the block descriptor of disks to indicate whether a disk is an instance storage disk would solve the issue.
We use the virtio block driver for disks attached to VSIs. One reliable method for achieving the above is through the virtio block request type of VIRTIO_BLK_T_GET_ID, which returns 20 bytes of identifier information. If IBM Cloud can change these 20 bytes of ID returned to a specific 20 bytes that indicate the device is an "instance storage disk" or "persistent disk", we can then identify and uniquely handle the device correctly at boot.
The above suggested method would provide a long-term reliable solution for identifying instance storage disks in IBM Cloud VSIs.
| Idea priority | Medium |
| Needed By | Not sure -- Just thought it was cool |
By clicking the "Post Comment" or "Submit Idea" button, you are agreeing to the IBM Ideas Portal Terms of Use.
Do not place IBM confidential, company confidential, or personal information into any field.