Skip to content
Tutorials Center
My learning

Topics

Lesson 3 of 45 ยท Cloud Basics

Regions, Zones and Latency

Understand Google Cloud regions, zones and multi-regions, how to choose a location for latency, price, availability and data residency, with a region picker.

Every Google Cloud resource lives somewhere physical. A region is a geographic area such as us-central1 (Iowa) or europe-west1 (Belgium). Inside a region are several zones, which are separate deployment areas with independent power, cooling and networking, named like us-central1-a. Where you place a resource affects speed, price, resilience and sometimes legal compliance.

Choosing a location is one of the few decisions that is hard to undo. A VM can be recreated elsewhere, but a database with terabytes of data, or a bucket, cannot simply be moved. This lesson explains the three kinds of location and a simple way to decide.

Regions, zones and multi-regions

Three kinds of location

Region A geographic area, like europe-west1.
Zone One failure domain inside a region, like europe-west1-b.
Multi-region A large area spanning regions, like US or EU.

Names follow region-then-letter, so the zone says its region.

1 / 2

A zone is the unit you pick when creating a virtual machine or a zonal disk. A region is the unit for many managed services such as Cloud Run, Cloud SQL and regional buckets. A few services, like Cloud Storage and BigQuery, also offer multi-region locations such as US and EU, which keep copies in more than one region for higher availability at a different price. Some resources (like VPC networks, global load balancers and IAM) are global and are not tied to one region.

The zone letter is just a label. Google may map a in your project to a different physical building than a in someone else's, so do not read meaning into the letters.

Terminal
# See the available regions and zones (needs a real account)
gcloud compute regions list
gcloud compute zones list --filter="region:europe-west1"

Zonal, regional and global resources

Every resource lives at one of three scopes, and the scope decides what a failure can take down and where you can use it.

Scope Examples
Zonal VM instances, persistent disks, Cloud TPUs, zonal managed instance groups
Regional Subnets, regional managed instance groups, Cloud Run services, Cloud SQL instances, Cloud Interconnect VLAN attachments, static regional IP addresses, regional buckets
Global VPC networks, firewall rules, global load balancers, global static IP addresses, images, snapshots, Pub/Sub topics
Multi-region Multi-region buckets, BigQuery datasets in US or EU, multi-region Spanner

Why you need to know: a VM can only attach disks from its own zone, so a zone outage stops the VM and its disks together. Per-zone activities such as maintenance events apply to VMs in that zone. To survive a zone failure you replicate across zones (a regional managed instance group, a Cloud SQL HA pair), and to survive a region failure you replicate sites and networks across regions (data replication, a second set of services, DNS or a global load balancer to steer traffic). Hybrid links such as Cloud Interconnect are made redundant by using attachments in more than one zone of the metro area (Hybrid Connectivity).

The need for regions and zones comes from three things: latency (put resources near users), fault isolation (failures stay inside one zone or region), and compliance (data must stay inside certain borders). Windows VMs in a zone are reached with RDP and Linux VMs with SSH, over the private network or through Identity-Aware Proxy.

Availability: why zones matter

Hardware fails and maintenance happens. If your whole application runs in one zone, an outage there takes you down. Typical patterns:

  • Single zone: cheapest and simplest; fine for experiments and things you can rebuild.
  • Multiple zones in one region: the common production default. A regional managed instance group or a high-availability database survives the loss of a zone. You will build this in Instance Groups, Autoscaling and Load Balancing.
  • Multiple regions: for the strictest availability needs, or users all over the world. It costs more and adds complexity such as data replication.

Google publishes availability targets (service level agreements) per product and configuration, and a multi-zone design usually earns a higher one than a single zone. Check the current SLA of the product you use rather than relying on a number remembered from a lesson.

How to choose a region

Weigh these factors, roughly in this order:

  1. Data residency and compliance. Laws or contracts may require data to stay in a country or economic area. Decide this first because it can rule options out.
  2. Latency to your users. Light in fiber is not instant: a round trip across an ocean adds tens to over a hundred milliseconds. Pick the region nearest the majority of users, or use a CDN for static content (Load Balancing, Cloud CDN and DNS).
  3. Latency between your own services. Keep the app, database and cache in the same region, and ideally close together, because chatty services multiply round-trip delays.
  4. Price. The same machine costs different amounts in different regions. Savings rarely outweigh latency or compliance, but matter for batch jobs that have no users waiting.
  5. Available products and capacity. Not every service or machine type exists in every region, and new features often launch in a few regions first.
  6. Carbon footprint. Google publishes how clean each region's energy is; some teams use it as a tiebreaker.

Also remember egress: data moving between zones, between regions or out to the internet may be charged. Keeping chatty components together saves both time and money. Exact rates change, so check the pricing pages.

Setting a default location

The gcloud CLI can remember a default region and zone so you do not repeat flags:

Terminal
gcloud config set compute/region europe-west1
gcloud config set compute/zone europe-west1-b

Many services have their own location flag (for example --region for Cloud Run), and a few set the location permanently at creation, so read the prompt before confirming.

A stand-in region picker

Real latency depends on the network path and must be measured. The numbers below are made up round figures so the logic is visible; the region names are real, but the latencies and prices are not Google's. The picker applies the decision order from above: filter by allowed locations first, then score the rest.

JS Pick a region by latency and price

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Notice how the residency filter ran before scoring. In real projects, a legal requirement is a hard constraint, while latency and price are trade-offs.

JS Zone names and availability after a failure

Press Run (or Ctrl + Enter) to see the console output here.

No output.

This only models the rule "two zones means one can fail". Real designs must also ensure that the remaining zone has enough capacity to take the whole load.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the relationship between a region and a zone?

    A region is a geographic area made up of multiple zones, which are separate failure domains inside it.
  2. 2. Which factor should you check first when choosing a region for customer data?

    Legal and contractual limits can rule regions out entirely, so they come before latency and price.
  3. 3. Your app runs in one zone and that zone fails. What would have protected you?

    Zones fail independently, so spreading copies across zones lets the others keep serving.
  4. 4. Why keep your app and its database in the same region?

    Every round trip between services adds delay, and moving data between regions can be charged.

Summary

  • A region is a geographic area; zones are independent failure domains inside it; multi-regions span several regions.
  • Spread across zones for resilience; use multiple regions for the strictest needs or global users.
  • Choose a location by residency rules, user latency, inter-service latency, price, product availability and sustainability.
  • Locations are hard to change later, so decide early.
  • Set default region and zone with gcloud config set.

Next: Account, Free Tier and Billing Setup