Mastering Azure VM Sizes and Disks: A Guide for AZ-104
Why This Topic is Critical
In Azure Infrastructure as a Service (IaaS), the selection of Virtual Machine (VM) sizes and disk types dictates not only the performance of your application but also your monthly cloud bill. Choosing the wrong size can lead to significant overspending or application failure due to resource starvation. Furthermore, specific availability Service Level Agreements (SLAs) are tied directly to the type of disk storage you select. Therefore, understanding how to deploy, manage, and resize these resources is a fundamental skill for any Azure Administrator and a high-priority topic on the AZ-104 exam.
What It Is
VM Sizes: Azure categorizes VMs into 'families' optimized for specific workloads. These define the number of vCPUs, amount of RAM, and temporary storage size.
General Purpose (D-series): Balanced CPU-to-memory ratio.
Compute Optimized (F-series): High CPU-to-memory ratio, ideal for traffic processing.
Memory Optimized (E-series): High memory-to-CPU ratio, ideal for relational databases.
Burstable (B-series): Low baseline performance with ability to burst; cost-effective for low-traffic workloads.
GPU (N-series): Specialized for graphics rendering and AI.
Azure Managed Disks: These are reliable, block-level storage volumes managed by Azure. Key types include:
Ultra Disk: Sub-millisecond latency, usage-based pricing, for mission-critical databases.
Premium SSD: High IOPS and throughput, required for the single-VM SLA (99.9%).
Standard SSD: Consistent performance at a lower cost, good for web servers.
Standard HDD: Magnetic backup storage, lowest cost, suitable for non-critical data.
How It Works
Resizing: When you change a VM size (Vertical Scaling), Azure attempts to resize the VM on the current hardware host. If the host supports the new size, the VM simply reboots. If the host hardware cannot support the new size, the VM must be deallocated (stopped) first, allowing Azure to migrate it to a different host cluster that supports the specifications.
Disk Administration: VMs come with an OS disk and a temporary disk (which is volatile—data is lost on reboot). Administrators attach persistent Data Disks for application storage. Disks can be expanded (increased in capacity) often without downtime, but they cannot be shrunk natively.
How to Answer Questions Regarding VM Sizes and Disks
Exam questions often present a scenario describing a workload and budget constraints, asking you to select the appropriate resource.
1. Identify the Constraint: If the question mentions 'minimize costs' for a test server, look for Standard HDD or B-series VMs. If the question mentions 'sub-millisecond latency,' the answer is Ultra Disk.
2. Check the Bottleneck: If a scenario states a VM typically runs at 10% CPU but crashes due to lack of RAM, you must switch to a Memory Optimized (E-series) size.
3. Understand SLAs: If a question asks specifically for a 99.9% SLA for a single VM, you must select Premium SSD or better. Standard SSD/HDD does not qualify for the single-instance connectivity SLA.
Exam Tips: Answering Questions on Manage VM sizes and disks
Memorize the Families: Remember the mnemonic mapping: B (Budget/Burstable), D (General Database/Daily use), E (Extra Memory), F (Fast Compute), N (Nvidia/GPU).
Resizing downtime: Always assume a resize requires a reboot. If a question asks how to resize a VM without ANY downtime, you generally cannot; you would need to deploy a second VM and load balance, rather than vertically scaling a live instance.
Disk Expansion vs. Shrinking: You can always make a disk larger (upsize). You cannot make a managed disk smaller. If a scenario asks to reduce a 1TB disk to 500GB to save money, the procedure is: Create new disk, migrate data, delete old disk.
Cached IOPS: Pay attention to 'Cached' vs 'Uncached' limits. Some VM sizes have limits on disk throughput. Even if you attach an Ultra Disk, if the VM size (e.g., a small B-series) has a low IOPS limit, the VM usually acts as the bottleneck, not the disk.