Simple explanation of persistence in Azure D-Series vs DS-Series

Question

I am looking at Azure D-Series Virtual Machines and comparing them with DS-Series Virtual Machines. I see the disk in D-Series is temporary SSD (whatever that means) while in DS-Series it's local SSD. My understanding so far was that temporary SSD means that live within that machine so if the machine changes (e.g. get scaled up or down) then the data will be lost. Here are my questions.

1. I created a (Linux) D-Series VM. I can see two disks. Why are there two and not one?
2. I placed some data in the "temporary" SSD and scaled the machine up and down. The data I originally created in that SSD are still there.
3. What are the performance differences - if any - between D-Series vs DS-Series machines?

Thanks Yannis

Answer 1

Yes, what are you talking about is the local SSD disk that is onboard by default. That local disk is not guaranteed to be persistent - so, Azure is talking about hardware failures that can delete the data. Or not. The same time, it is not guaranteed that it will be lost because it is localized to the local hardware.

Every VM has the persistent disk as well that is localized in the Azure Storage - that should be persisted whatever happens, because there is a replication, etc.

DS-series can use the Premium Storage which is the powerful storage. You can compare them by yourself (i would not include the data here because it may change in a future). But in a nutshell, D-series VMs are designed to run apps that demand higher compute power and temporary disk performance. DS-series VMs are able use Premium Storage, that provides high-performance, low-latency storage for I/O intensive scenarios.