27 regions,
36 availability zones

A region is a geographic area, usually one city, made up of one or more physically isolated zones with independent power; zones in a region are connected privately with latency under 2 ms, which is what high-availability deployments rely on.

Not by default. Connect them with the dedicated line product or go over the public internet; cross-region data sync is best done at the application layer or through database replication.

An instance itself cannot move between regions. Turn the system disk into a custom image, copy it to the target region and create a new instance from it; clone data disks from snapshots.

Ask sales to request stock, or deploy in a large region of the same area (Hong Kong, Singapore, Frankfurt, Washington), where public latency is usually under 30 ms.

Every region provides a latency test address (see the console); test with ping or mtr from where your users are, or spin up the smallest instance by the hour for a real load test and release it.

They are planned around customer demand; if you need a deployment in a specific country, register the need with sales — regions open first where demand reaches scale.