What it is
Multi-AZ means the important pieces actually run in more than one Availability Zone. Not “we could.” Running. Replicated. Able to fail over without a hero on a laptop.
A database with a standby in a second zone is multi-AZ. A web farm with instances in one zone and a nice drawing of two is not.
AWS RDS and Azure’s zone-redundant options charge extra for this. That extra is the price of not betting the company on one building’s air conditioning.
Why it matters in the meeting
This is the follow-up to Availability Zones. Zones are the buildings. Multi-AZ is whether you used more than one.
If the architecture slide says multi-AZ and the database console says a single zone, the slide is fiction. Bart will open the console.
Real world
When a zone fails, single-AZ workloads go to the incident channel. Multi-AZ workloads get a dull notification and keep serving customers who do not care about your diagram.
Storage, compute, and data each need their own answer. A multi-AZ load balancer in front of a single-AZ database is a very polite way to fail more slowly.
Multi-AZ is not a feature you turn on with a vibe. It is copies in two buildings, plus a plan for which copy takes over. If only one copy exists, you are single-AZ with better stationery.
What to ask
- Is the production database zone-redundant right now, in the console, not in the deck?
- Are compute instances actually spread across zones, or auto scaling in a single subnet?
- What is the failover time we have tested, not the one in the vendor brochure?
- Does “multi-AZ” apply to AWS, Azure, or only the cloud we talked about last Tuesday?
You just knew a little more Jack than you did five minutes ago.
All concepts