Today, we're bringing Google Compute Engine customers two highly requested features for Persistent Disks. First is the ability to create and delete root persistent disk volumes in the same command that creates and deletes VMs. The second is an easier way to grow a Persistent Disk volume. In addition, we are increasing the maximum IOs per second that a VM can issue to a Persistent Disk volume by 50% on reads and by over 500% on writes.

Create and delete a root Persistent Disk with an instance
Many customers appreciated the convenience of being able to create or delete a VM instance with a single API call that scratch disk provided, and we've just added that single-API simplicity for creating or deleting VM instances that use root persistent disks. Check out Starting an Instance in the API and Setting the Auto-delete State of a Root Persistent Disk for full details.

Resizing Persistent Disks
Customers want to be able to start with a conservative size and IO limit and then grow that volume as the data and IO needs increase. They also want to be able to do this without having to migrate all the data to a new volume.

As of today, that is now possible. Following the instructions here, you can now snapshot a volume and restore that snapshot to a volume with a larger size. That’s it! The larger volume automatically gets higher IO caps because we reserve more storage for that volume. IO caps are described here. Please note that you will need to extend the partition and filesystem inside the VM as described in the instructions to make use of the extra space.

This feature allows you to be more conservative with your Persistent Disk provisioning and allows you to quickly and painlessly grow the volume later if you find you need more space or IO. Try it out and let us know how it works for you.

Increasing VM Performance To Persistent Disk
Our persistent disk offering is unique because developers can get a high level of performance with low cost block storage. Persistent Disk volumes can perform up to 0.3 read IOPS and 1.5 write IOPS for each GB of allocated space. So, a 1 TB volume can do up to 300 read IOPS and 1500 write IOPS. Previously, there was an additional IOPS cap at the VM level of 2000 read IOPS and 2400 write IOPS regardless of volume size. Those caps have been removed for all VM sizes. Larger instance types with greater IO capability can now see persistent disk performance grow linearly all the way up to the maximum volume size of 10TB. For example, an 8 core VM with a 10TB volume now will be able to execute up to 3000 read IOPS or 15000 write IOPS all still at the price of $0.04/GB/month with no extra IO charges.

-Posted By Jay Judkowitz, Product Manager