How to calculate total availability based on individual object availabilities?

Question

How to calculate overall availability given individual object availability percentages?

Example 1: I have sql server with FCI.

VM1 has 99.9% availability

VM2 has 99.9% availability

What is total availability calculation?

Example 2: same as above plus disk storage has 99.9% availability. So what is total availability?

Example 3: Suppose I have 1 VM and 1 disk storage. Each with 99.9% availability. So what is total availability calculation?

Answer 1

AWS is using Calculating Total System Availability article for availability calculation. So in your setup you have to recognize what is parallel and what is serial relationships. For basic scenarios:

Serial Configuration

A(VM1 + disk) = 0.999 * 0.999 = 0.998

Parallel Configuration

Assuming VM2 is a redundancy for VM1

A(VM1, VM2) = 1 - (1-0.998)(1-0.998) = 0.99996

Answer 2

Example 1: If you have 2 VMs and each one has 99.9% availability and your total availability like below

.999 * .999 = .998 it means monthly 1h 27m 39s of downtime.

Example 2: If you have 2 VMs and each one has 99.9% availability and one more disk storage has 99.9% availability. your total availability like below

.999 * .999* .999=.997 it means monthly 2h 11m 29s of downtime.

Example 3: if you have 1 VM and 1 disk storage. Each with 99.9% availability and your total availability like below

.999* .999=.998 it means monthly 1h 27m 39s of downtime.

The SLA for Google cloud VMs when you have two or more VMs configured across various availability zones is 99.99%.

You can also use this link for compute engine and storage availability and use link for SLA calculations,