MultiCloud Classroom notes 29/Jul/2025

AWS Instance Types by Purpose

• General Purpose: Balanced CPU to RAM Ratio. This is good at a general purpose level
• CPU Optimized: High CPU to RAM Ratio.
• Memory Optimized: High RAM to CPU Ratio
• Storage Optimized: Better Disk speeds
• Accelerated Computing
• HPC
• Instance Families: Each instance family represents hardware specifications
  • Generally an alphabet represents it t in t2.micro
  • Instance family will have generations t2 & t3
  • In each family & generation we have instance types
• Instance type:
  • vCPU
  • RAM
  • Network Performance
• Instance type Pattern / naming conventions
• Example-1

c7gn.2xlarge
# Summary
# Compute Optimized
# Generation 7
# g – AWS Graviton processors
# n – Network and EBS optimized
# size = 2xlarge

• Example 2

z1d.xlarge

# Summary
# Z – High memory generation 1
# d – Instance store volumes

Azure Vm Sizes by Purpose

• General Purpose: Balanced CPU to RAM Ratio. This is good at a general purpose level
• CPU Optimized: High CPU to RAM Ratio.
• Memory Optimized: High RAM to CPU Ratio
• Storage Optimized: Better Disk speeds
• GPU Accelerated
• FPGA Accelerated.
• HPC
• VM Family: Each VM family represents hardware specifications
  • Generally an alphabet represents it B in B1s
  • In each family & Generation we have sizes supported
• Instance type Pattern / Naming convention

M416ms_v2

# Summary 
# M Family
# 416 vCPUS
# m = memory intensive; the most amount of memory in a particular size
# s = Premium Storage capable, including possible use of Ultra SSD (Note: some newer sizes without the attribute of s can still support Premium Storage, such as M128, M64, etc.)
# Version = v2

• Example 2

E96as_v4

# Summary
# E family
# 96 vcpus
# a = AMD-based processor
# s = Premium Storage capable, including possible use of Ultra SSD (Note: some newer sizes without the attribute of s can still support Premium Storage, such as M128, M64, etc.)
# Verison v4

Decision Tree: Selecting Azure VM Size

1. Identify Workload Type
2. General Purpose (D, B, Av2-series): For common workloads (web/app servers, small-medium databases)
3. Compute Optimized (F, Fs-series): For compute-intensive tasks (batch, analytics, gaming)
4. Memory Optimized (E, M-series): For large in-memory DBs, data warehouses, memory-heavy apps
5. Storage Optimized (L-series): For high disk I/O needs (NoSQL, big data)
6. GPU Optimized (N-series): For graphics, ML, deep learning workloads
7. High Performance Compute (H, HB, HC-series): For demanding scientific/engineering tasks
8. Determine Core Resource Needs
9. vCPUs: Evaluate required processing power
10. RAM: Estimate memory footprint
11. Disk I/O (IOPS/Throughput): Assess read/write storage needs
12. Network Performance: Identify necessary bandwidth
13. Consider Specific Features
14. Premium Storage Support: Needed for high-performance SSDs?
15. Accelerated Networking: Required for low latency/high throughput?
16. Confidential Computing: Necessary for sensitive data processing?
17. Processor Architecture: Preference for Intel or AMD CPUs?
18. Evaluate Cost
19. Compare VM sizes within your chosen series for cost-effectiveness
20. Consider Azure Reserved VM Instances for long-term savings
21. Refine and Test
22. Select a candidate VM size based on previous steps
23. Deploy workload and monitor actual performance
24. Adjust VM size as necessary for performance and cost

Example Table (VM Series vs. Workload Type)

| Workload Type | Recommended Series | Notes |
|———————— |———————- |———————————————————————————|
| General Purpose | D, B, Av2 | Balanced CPU:RAM, most common workloads |
| Compute Optimized | F, Fs | High CPU requirements |
| Memory Optimized | E, M | High memory needs |
| Storage Optimized | L | High disk throughput needs |
| GPU Optimized | N | ML, graphics, and visualization |
| High Performance Compute| H, HB, HC | Scientific, engineering, high-performance needs |