Examkin

Question

Difficulty: Very hardDesign Compute High Availability

A financial services corporation is architecting a real-time transaction processing platform. The core processing tier will run on Azure Virtual Machines. The platform has the following infrastructure requirements:
- A virtual machine uptime Service Level Agreement (SVA) of 99.99%.
- Under 1 millisecond network latency between the active virtual machine nodes in each cluster partition to support memory-mapped replication.
- Complete protection against a localized datacenter power or cooling failure.

Which design should you recommend to meet these requirements?

  1. Distribute the virtual machine instances across multiple Availability Zones using a Virtual Machine Scale Set in Flexible orchestration mode. Create a separate Proximity Placement Group for each Availability Zone and associate the virtual machines in each zone with their respective Proximity Placement Group.Answer
  2. B
    Group all virtual machine instances within a single Availability Zone using a single Proximity Placement Group, and place the instances in an Availability Set.
  3. C
    Configure the virtual machine instances as Azure Spot Virtual Machines within a Virtual Machine Scale Set in Flexible orchestration mode, spanning multiple Availability Zones with a single Proximity Placement Group.
  4. D
    Span the virtual machine instances across multiple Availability Zones using a single Proximity Placement Group that is shared by all instances across all zones.

Answer

Distribute the virtual machine instances across multiple Availability Zones using a Virtual Machine Scale Set in Flexible orchestration mode. Create a separate Proximity Placement Group for each Availability Zone and associate the virtual machines in each zone with their respective Proximity Placement Group.
Distributing the virtual machines across multiple Availability Zones provides the necessary 99.99% VM uptime SLA and zone-redundancy. Using a separate Proximity Placement Group (PPG) for each zone ensures that the virtual machines within each zone are colocated for ultra-low latency, while avoiding deployment allocation constraints that occur when stretching a single PPG across multiple zones.

Step-by-Step Solution

1
Determine the SLA requirement and select the correct placement structure.
To meet a 99.99% VM uptime SLA, the virtual machines must be distributed across two or more Availability Zones.
Azure provides a 99.99% VM uptime SLA only when VMs are deployed across multiple Availability Zones in the same region.
2
Address the low-latency requirement within each cluster partition.
A Proximity Placement Group (PPG) is required to group VMs closely together physically to achieve sub-millisecond network latency.
A PPG is a logical grouping capability used to make sure Azure compute resources are physically located close to each other.
3
Analyze constraints when combining Availability Zones and PPGs.
Creating a single PPG that spans multiple Availability Zones is not recommended and often fails because Azure cannot satisfy both physical zone separation (for high availability) and close physical proximity (for the PPG).
Stretching a single PPG across zones results in deployment allocation failures due to conflicting physical constraints.
4
Formulate the final design.
The optimal configuration is to deploy the virtual machines across multiple zones using a Virtual Machine Scale Set (VMSS) in Flexible orchestration mode, and deploy a separate PPG within each Availability Zone.
This configuration satisfies the 99.99% SLA, guarantees low latency within each zone, and avoids placement constraint errors.

Key Concept

Interaction between Azure Availability Zones, Proximity Placement Groups, and VMSS orchestration modes to balance high availability and low latency.
Rate this question

Topics

Design Compute High Availability

More questions from this topic