This guide is for CTOs and IT leads about to run production workloads in a public cloud. It sits under Netbase's cloud platform engineering work, and it is the step that comes before a cloud migration moves anything.
In this guide
- What does a minimal landing zone have to contain?
- Which accounts, subscriptions or projects should exist on day one?
- What network decisions are hard to change later?
- Which guardrails should be switched on first?
- Accelerator, custom build or single account: which start fits?
- The minimal landing-zone checklist in build order
- When should you not build a full landing zone?
- Where AI fits in landing-zone work
- What delivery record exists, and what does not
- Alternatives: other ways to get a governed cloud foundation
- What are the limits of this guide?
- Frequently asked questions
- Next step
What does a minimal landing zone have to contain?
A landing zone is the governed foundation that workloads land in: identity, an account or subscription structure, guardrails, logging and a network layout, ideally written as code. It is built once, before the first production workload, and extended as new teams arrive.
AWS defines a landing zone as a baseline for multi-account architecture, identity and access management, governance, data security, network design and logging, built either with AWS Control Tower or as a customised landing zone. Google Cloud lists four design areas that every landing zone must settle: identity provisioning, resource hierarchy, network design and security controls. Google also notes that the first iteration is often not the final one.
Microsoft splits an Azure landing zone into one platform landing zone, which holds governance and shared services, and workload landing zones, where application teams deploy inside the platform's guardrails. Most organisations need only one platform landing zone per Microsoft Entra tenant.
The three providers use different words for the same building blocks:
-
Isolation unit
- AWS
- Account
- Azure
- Subscription
- Google Cloud
- Project
-
Grouping for policy
- AWS
- Organizational unit (OU)
- Azure
- Management group
- Google Cloud
- Folder
-
Top of the hierarchy
- AWS
- Organization with a management account
- Azure
- Tenant root management group
- Google Cloud
- Organization node
-
Preventive guardrail
- AWS
- Service control policy (SCP)
- Azure
- Azure Policy with deny effect
- Google Cloud
- Organization policy constraint
-
Starter tooling
- AWS
- AWS Control Tower
- Azure
- Azure landing zone accelerators
- Google Cloud
- Terraform example foundation or console setup
Which accounts, subscriptions or projects should exist on day one?
Start with fewer units than an enterprise reference design, but never with one. The day-one minimum is a management or root unit that runs no workloads, a log archive, a security tooling unit, and one production and one non-production unit per workload.
AWS explains why in its guidance on organising an environment with multiple accounts: an account is an isolation boundary for identity and access, no access is allowed between accounts by default, and charges follow resource usage rather than the number of accounts. The same paper says a team experimenting in a single account does not need these practices until it plans its first few production workloads.
Keep the management account empty. AWS service control policies do not affect users or roles in the management account, so anything running there sits outside every guardrail you set. On Azure, keep platform subscriptions under their own management groups, apart from workloads.
-
Management or root
- Purpose
- Billing, organisation settings, nothing else
- Can wait
- Never skip
-
Log archive
- Purpose
- Central, write-protected audit and access logs
- Can wait
- Never skip
-
Security tooling
- Purpose
- Security findings and break-glass access
- Can wait
- Never skip
-
Workload production
- Purpose
- The first real application
- Can wait
- Never skip
-
Workload non-production
- Purpose
- Test and staging for the same application
- Can wait
- Never skip
-
Shared network hub
- Purpose
- Central egress, firewall, hybrid links
- Can wait
- Until a second workload or on-premises link needs it
-
Sandbox
- Purpose
- Free experiments with a budget cap
- Can wait
- Until engineers ask for one
-
Account or subscription vending
- Purpose
- Automated creation of new units
- Can wait
- Until requests outgrow a manual, scripted process
What network decisions are hard to change later?
The IP address plan is the decision that is hardest to reverse. Allocate non-overlapping ranges per environment and region before the first virtual network exists, leaving room for on-premises networks, partner links and regions you do not use yet.
Three other choices are worth making on paper before day one, even if they are built later: hub-and-spoke or a managed WAN topology, how private DNS resolves across units, and where internet egress leaves the estate. Microsoft's reference architectures show both a hub-and-spoke and a Virtual WAN layout for exactly this reason: the topology shapes every workload network that follows.
What can wait: a firewall appliance, cross-region routing and a private office link, which Google says can come later. Application teams then check their VPC design.
Which guardrails should be switched on first?
Switch on a short list of preventive guardrails at the top of the hierarchy and test them on a non-production unit first. AWS strongly recommends not attaching an SCP to the organisation root before testing it, and suggests moving accounts into a test OU one at a time so nobody is locked out of key services.
A minimal first set covers six controls: deny use of unapproved regions, block public access to storage by default, protect the central audit trail from changes, deny long-lived root or owner credentials, require tags for owner and cost centre, and send budget alerts per unit. Google's Organization Policy Service can restrict where new resources are created, policies are inherited by every folder and project below, and a dry-run mode logs violations without denying them.
Guardrails limit, they do not grant. AWS states that an SCP never grants permissions; teams still need identity policies that give them access, so plan the permission sets alongside the guardrails rather than after them.
Accelerator, custom build or single account: which start fits?
| Starting point | Strength | Watch out for | Choose it when |
|---|---|---|---|
| Provider accelerator (Control Tower, Azure accelerators, Google foundation) | Follows the provider's recommended practice; fastest path | Opinionated layout and tooling your team must learn | You run mainly on one cloud and accept its defaults |
| Custom modules in your own infrastructure as code | Fits existing tooling, naming and review flow | You own every control and every upgrade | Your team already runs infrastructure as code well |
| Single account with tags and budgets | Quick to start, little to learn | Weak isolation; moving later means migrating resources | A short pilot with no production data |
Microsoft states that for most organisations its accelerators give the fastest path to a deployment aligned with recommended practice. The trade-off is fit: a team that already reviews every change in Terraform may lose time adapting to an accelerator's tooling.
The minimal landing-zone checklist in build order
-
Connect identity
Federate one workforce identity provider, require multi-factor sign-in and lock away root or owner credentials.
-
Create the hierarchy
Set up the organisation, the platform and workload groups, and the day-one units from the table above.
-
Centralise logs
Send audit and access logs from every unit to the log archive and protect them from changes.
-
Apply guardrails
Attach the first preventive policies to a test unit, check nothing breaks, then move them up the hierarchy.
-
Plan the network
Write the IP address plan, the DNS approach and the egress route; build only what the first workload needs.
-
Set cost controls
Add required tags, budgets and alerts per unit, and review commitment discounts only once usage is stable.
-
Put it in code
Keep the whole foundation in a repository with reviewed changes and a pipeline, so the next unit is created the same way.
When should you not build a full landing zone?
Do not build an enterprise-scale hierarchy for one application and one team; the minimal set above is enough until a second team or workload arrives. A short pilot with no production data can start in one sandbox account with budgets, if everyone agrees it will be thrown away.
Do not copy a reference hierarchy you cannot staff: Microsoft advises centralising only capabilities with a clear governance, operational or economic benefit across several workloads.
Where AI fits in landing-zone work
AI helps with the reading and checking around the foundation, not with the decisions. An assistant can draft infrastructure-as-code modules from a written checklist, compare a policy set against provider documentation and summarise policy violations for review. Netbase works with the major commercial and open-source AI tools and models, chosen per project.
Engineers still write the address plan, decide every guardrail, review every generated module line by line and apply changes through the pipeline; no model is given rights to change the organisation, its policies or its networks.
- In delivered work
MLOps pipeline
An MLOps pipeline Netbase delivered for a client that is not named; an AI delivery record, not a landing zone.
- Available capability
AI agents in operations
AI-assisted policy review and module drafting is a way Netbase approaches the work, not yet tied to a published case.
What delivery record exists, and what does not
Netbase names AWS, Google Cloud, DigitalOcean and Cloudflare as the cloud providers it uses. It also works as offshore development partner on a multi-tenant cloud ERP delivered as SaaS to small and mid-sized businesses for a US client, not named.
Delivery runs in a six-step lifecycle with weekly reviews and KPI dashboards, and most Netbase projects are delivered on fixed-price contracts. No published Netbase record describes a landing-zone build in detail, no Azure project is claimed, and no duration, account count or result figure is given here.
Alternatives: other ways to get a governed cloud foundation
| Alternative | Fits when | Selection criterion |
|---|---|---|
| Provider accelerator as-is | One main cloud, standard needs | You accept the provider's layout and tooling |
| Accelerator plus your own modules | Standard core, a few local rules | Your extra controls can sit on top without forking it |
| Fully custom foundation | Strong infrastructure-as-code team, several clouds | You can own and upgrade every control yourself |
| Managed platform team or partner | No in-house platform engineers yet | The partner hands over code, documentation and access |
Once the landing zone is in place, the move itself follows a cutover plan with rollback points, and pipelines and environments are the work of Netbase's DevOps and platform engineering service.
What are the limits of this guide?
- It is a minimal checklist, not a compliance design; regulated workloads add controls on top of it.
- Provider tools, names and defaults change; check the current documentation of your provider before you build.
Plan the next step with a Netbase consultant
Frequently asked questions
A cloud landing zone is the governed foundation that workloads are deployed into: identity, an account or subscription structure, guardrails, central logging and a network layout. It is set up before the first production workload and extended as new teams and workloads arrive.
You need a small one. One application still benefits from separate production, non-production, log and security units and a few guardrails, but it does not need a network hub, account vending or a large group hierarchy until a second team or workload arrives.
Use the provider's tooling when you run mainly on one cloud and accept its defaults; build your own modules when your team already runs infrastructure as code well.
The IP address plan. Overlapping ranges block peering and private links later, and fixing them means rebuilding networks under running workloads, so the plan is written before the first virtual network exists.
Next step
If you want a review of your landing-zone plan before the first workload moves, book a solution review with Netbase.
Related services and solutions
AI-ready cloud platform engineering for SaaS and commerce workloads Netbase provides cloud platform engineering for SaaS and commerce teams that need a secure, repeatable home for their software and its AI features, on AWS, Google Cloud, DigitalOcean or Cloudflare. We design the architecture, write the environments as code, connect managed and AI services and hand over a platform your team can run, built the way we run ours.
Learn More
AI-assisted cloud migration with a cut-over plan and a way back Netbase provides cloud migration services for companies running business applications on ageing servers or hosting to move them to AWS, Google Cloud, DigitalOcean or Cloudflare without losing data or control. AI speeds up the code and dependency assessment; every migration still has an inventory, a rehearsed cut-over, cost controls from day one and a tested rollback point.
Learn More
AI-assisted DevOps and platform engineering for routine releases Netbase provides DevOps and platform engineering for SaaS and commerce teams to make releases routine instead of risky. We set up CI/CD pipelines, repeatable environments and observability on AWS, Google Cloud, DigitalOcean or Cloudflare, with AI assistance for tests, reviews and incident triage, so every change ships the same way and problems surface before customers notice.
Learn More
SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules A SaaS product accelerator is a set of reusable Netbase modules for accounts, billing, roles and integrations that helps founders and product teams launch subscription software faster, with room for in-product AI from the first release. Reusing these modules can shorten development time, and the approach is proven on Printcart, Netbase's own web-to-print SaaS.
Learn More
Work with Netbase remotely
Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.project@netbasejsc.com
+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.