study guide
Orientation

Start here

This guide takes you from "what is a landing zone?" to being ready to brief a customer and run a Landing Zone Accelerator (LZA) deployment with the Universal Configuration. No prior AWS multi-account experience is assumed.

By the end you will be able to

  • Explain what a landing zone is and why customers need one
  • Describe what LZA is, what it is not, and how it relates to AWS Control Tower
  • Walk through how LZA is installed and how its pipelines work
  • Name the 8 configuration files and what each one controls
  • Describe every account, OU, and guardrail the Universal Configuration creates
  • Trace network traffic through the hub-and-spoke design
  • Explain the security, logging and backup services that get turned on
  • Collect the right inputs from a customer before deploying
  • Handle common customer questions and objections
  • Operate the environment after day one: add accounts, upgrade, troubleshoot

How to use this guide

1. Read in order

Modules build on each other. Each ends with a Mark complete button so the sidebar tracks your progress (saved in this browser).

2. Click everything

Diagrams are interactive: click accounts, VPCs, pipeline stages and pools to see details. Play the traffic flows in the network module.

3. Test yourself

Finish with the flash cards and the quiz. Aim for 85%+ on the quiz before your first customer briefing.

Suggested learning path

SessionModulesGoalTime
1Landing zones & LZA, AWS building blocksSpeak the vocabulary45 min
2How LZA works, Deploying LZA (with the interactive walkthrough)Understand the machinery60 min
3Universal Configuration: overview, accounts & guardrailsKnow what gets built45 min
4UC networking, UC security & loggingExplain the architecture60 min
5Day 2 operations, Briefing a customerBe customer-ready45 min
6Flash cards, Quiz, GlossaryLock it in30 min
Versions coveredLZA v1.17.0 (latest release at the time of writing) and Universal Configuration v1.4.0 (released November 4, 2026). Always check the release notes before a real deployment: both projects ship often.
ImportantLZA does not, by itself, make a customer compliant with any framework. It builds a strong, well-architected foundation; the customer still has to review, assess and approve it against their own requirements. Say this early in every briefing.
Foundations · Module 1

What is a landing zone, and what is LZA?

Before any technology: the problem LZA solves, in plain language.

The problem: one AWS account doesn't scale

A brand-new organization often starts with a single AWS account. Everything lives together: development experiments, production databases, security logs. That becomes a problem quickly:

  • Blast radius. A mistake (or an attacker) in one workload can affect everything else in the account.
  • Permissions sprawl. It's hard to give the dev team freedom without also giving them access to production.
  • Billing confusion. You can't easily tell which team or project spent what.
  • Compliance. Auditors want logs that nobody can delete, and evidence that controls are applied everywhere.

The AWS best practice is to use many accounts: one per environment, team or workload, grouped and governed centrally.

AnalogyThink of an AWS account as an apartment. One big open-plan apartment for everyone is chaotic. A landing zone is the whole apartment building: separate units (accounts), floors for different purposes (organizational units), building rules everyone must follow (guardrails), a security desk with cameras (logging and security services), and shared plumbing and electricity (networking).

What is a landing zone?

A landing zone is a pre-built, secure, multi-account AWS environment, ready for teams to "land" their workloads in. A good one includes:

Account structure

Accounts grouped into organizational units (OUs) by purpose: security, infrastructure, workloads.

Identity

One place where people sign in (single sign-on) and get the right access to the right accounts.

Guardrails

Rules that prevent risky actions (preventive) and spot misconfigurations (detective).

Central logging

Every API call and important log copied to an account that nobody can tamper with.

Security services

Threat detection, posture checks and findings, managed centrally.

Networking

Shared, inspected connectivity between accounts, to the internet, and to on-premises.

What is the Landing Zone Accelerator on AWS?

LZA is a free, open-source AWS solution that builds and maintains a landing zone from configuration files. You describe what you want in YAML text files, and LZA's pipeline deploys it across all your accounts and Regions, and keeps it that way.

What it is

  • An AWS Solution, published by AWS, open source on GitHub (awslabs/landing-zone-accelerator-on-aws)
  • Built with the AWS Cloud Development Kit (CDK), deployed through AWS CodePipeline
  • Driven by 8 YAML configuration files, not code you write
  • Covers 35+ AWS services and features
  • No additional charge: you pay only for the AWS services it turns on
  • Works in commercial Regions, AWS GovCloud (US), and other partitions

What it is not

  • Not a compliance certificate: it helps you align with frameworks
  • Not a managed service: the customer owns and runs it
  • Not an application platform: it builds the foundation, not workloads (for example it doesn't deploy your load balancers or apps)
  • Not a replacement for Control Tower: it is designed to work with Control Tower

LZA and AWS Control Tower

AWS Control Tower is an AWS service that sets up a basic landing zone: an organization, a Security OU with Log Archive and Audit accounts, an organization CloudTrail, IAM Identity Center, and a catalog of controls. AWS recommends Control Tower as the foundation and LZA on top to extend it.

CapabilityControl Tower aloneControl Tower + LZA
Organization, OUs, Log Archive and Audit accounts✓✓ (LZA can even set up Control Tower for you)
Controls (guardrails) from the Control Tower catalog✓✓ plus custom SCPs, RCPs, declarative, tag and backup policies
Centralized networking (Transit Gateway, firewall, IPAM, endpoints)—✓
Security services org-wide (GuardDuty, Security Hub, Macie, Config rules)Partial✓ configured and delegated
Everything as version-controlled configuration with a deployment pipeline—✓
Drift detection and automatic re-applicationControl Tower drift only✓ every pipeline run re-applies the config

Related tools you may hear about: Customizations for Control Tower (CfCT) and Account Factory for Terraform (AFT) also extend Control Tower. LZA is the most complete "configuration-file driven" option, especially for regulated customers.

Who uses LZA?

LZA was designed for customers with strict security and compliance needs: public sector, defense, healthcare, financial services, and anyone who wants a well-architected foundation fast. AWS publishes best-practice configurations for it, and the Universal Configuration (covered later) is the recommended general-purpose starting point.

RememberLanding zone = the secure multi-account building. LZA = the robot that builds and maintains it from config files. Control Tower = the foundation LZA builds on. Universal Configuration = a ready-made, best-practice set of those config files.
Foundations · Module 2

The AWS building blocks you need to know

LZA wires together dozens of AWS services. These are the ones you must be able to explain simply. Click a category to filter.

TipEvery one of these terms is also a flash card and in the glossary. If a customer asks "what's an SCP?", your answer should fit in two sentences.
LZA · Module 3

How LZA works under the hood

Two pipelines, one set of configuration files, and a lot of CloudFormation. Once you understand this loop, everything else makes sense.

The big picture

  1. You launch one CloudFormation template (AWSAccelerator-InstallerStack) in the management account. It takes about 8 minutes.
  2. The Installer pipeline (AWSAccelerator-Installer) downloads the LZA source code from GitHub and deploys the second pipeline. Its only job is to install or upgrade LZA itself.
  3. The Core pipeline (AWSAccelerator-Pipeline) reads your configuration files and deploys everything they describe, account by account and Region by Region, using AWS CDK and CloudFormation. The first run takes about 45 minutes and deploys a minimal baseline.
  4. You edit the configuration files and push the change to the configuration repository. The Core pipeline runs again and makes the environment match.
AnalogyThe configuration files are the architect's blueprint. The Core pipeline is the construction crew that builds exactly what the blueprint says, and rebuilds anything that doesn't match on every visit. The Installer pipeline is the company that hires (and upgrades) the crew.

Why two pipelines?

They are separate by design. Upgrading LZA to a new version only means changing one parameter (the release branch name, such as release/v1.17.0) on the installer CloudFormation stack. The Installer pipeline then rebuilds the Core pipeline with the new code, and the Core pipeline re-applies your unchanged configuration.

Inside the Core pipeline

Click each stage to see what it does. Stages run in order; if one fails, the pipeline stops there and you fix the cause and retry.

Where the configuration lives

When you launch the installer, the Configuration Repository Location parameter decides where your config files are stored. You cannot change it later without stack failures, so decide up front:

Amazon S3

The files live in an S3 bucket (aws-accelerator-config-…) as a zip. Simple; you upload a new zip to deploy.

AWS CodeCommit

A Git repository in the management account. The historical default. Check CodeCommit availability for new customers.

CodeConnections

A Git repository you already use (for example GitHub or GitLab) connected through AWS CodeConnections. Best for teams that review changes with pull requests.

The 8 configuration files

Every LZA environment is described by these YAML files, plus folders of supporting files (policy JSON, firewall rules, SSM documents). Click a file.

Replacements: variables for your config

replacements-config.yaml defines named values that the other files reference with double curly braces. LZA substitutes them before reading the files. Change a value once, and it updates everywhere.

replacements-config.yaml

globalReplacements:
  - key: HomeRegion
    type: String
    value: us-east-1
  - key: GlobalCidr
    type: String
    value: "10.0.0.0/8"

network-config.yaml

transitGateways:
  - name: "{{ AcceleratorPrefix }}-{{ HomeRegion }}-tgw"
    region: "{{ HomeRegion }}"
# becomes: AWSAccelerator-us-east-1-tgw

Policy files can also use tokens that LZA fills in at deploy time, such as ${ORG_ID}, ${PARTITION} and ${ACCELERATOR_PREFIX}.

Three behaviors to remember

Validation first

The Build stage checks your files against a schema. Typos and invalid values fail fast, before anything changes.

Desired state

Every run re-applies the full configuration. Manual console changes to LZA-managed resources get overwritten (and many are blocked by SCPs).

Optional approval

With Enable Approval Stage = Yes, the pipeline pauses after a CDK diff so someone can review and approve changes. Recommended for any landing zone that runs real workloads: one pipeline run deploys to every account at once, Dev and Prod alike.

One landing zone, many environmentsAn LZA installation manages one landing zone: one AWS organization with one pipeline and one set of config files. Your Dev, Test, Prod and Sandbox environments live inside it as OUs and accounts, and every pipeline run updates all of them together. There is no "Dev pipeline" and "Prod pipeline".

Because a config change hits Prod the same moment it hits Dev, some customers build a second, separate landing zone (its own AWS organization and its own LZA install) only to try out config changes and LZA upgrades safely before applying them to the real one. When this guide says "test in a non-production landing zone", it means that separate organization, not the Dev OU.
RememberInstaller stack → Installer pipeline → Core pipeline → your config files → CloudFormation stacks in every account and Region. Change the files, not the resources.
LZA · Module 4

Deploying LZA, step by step

The full journey from a brand-new AWS account to a running landing zone, with and without Control Tower. Use the interactive walkthrough below, then the checklist.

Phases at a glance

PhaseWhat you doWhat happensTime
0. PlanCollect inputs: emails, home Region, Regions, IP ranges, identity provider, network modelNothing in AWS yetDays–weeks (customer decisions)
1. PrepareCreate the management account, enable AWS Organizations (all features), raise quotas if neededAn empty organization~1 hour
2. LaunchDeploy AWSAccelerator-InstallerStack with your parametersInstaller pipeline, then Core pipeline~8 min stack + pipelines
3. AwaitWatch the first Core pipeline runControl Tower (if enabled), Log Archive and Audit accounts, keys, log buckets~45 min (longer with Control Tower setup)
4. ConfigureCopy in the Universal Configuration, update emails, Region, CIDRsNothing until you pushHours
5. Deploy configPush to the configuration repositoryAccounts, OUs, guardrails, security services, networking1–several hours
6. ValidateCheck the pipeline, sign in through Identity Center, run the Controls Test PackEvidence the controls workHours

Interactive walkthrough

Three tutorials in one: building a landing zone with Control Tower, without Control Tower, and what happens inside one pipeline run. Click through each step and watch the diagram grow.

Control Tower prerequisites (when LZA sets it up for you)

The Universal Configuration expects Control Tower. LZA can deploy Control Tower itself if the organization is clean:

If the customer already has Control Tower, LZA can work with it, but plan carefully. The Universal Configuration docs recommend starting with a new organization for the best experience.

Do I need to bootstrap the management account?No. You don't run cdk bootstrap anywhere yourself. The Installer pipeline bootstraps the management account automatically (in the home Region and the global Region, such as us-east-1) with a stack named AWSAccelerator-CDKToolkit. The Core pipeline's Bootstrap stage then does the same for every other account and Region in your configuration. If the customer later deploys their own CDK apps, they should use their own bootstrap template so it doesn't collide with LZA's.

Installer stack parameters that matter

ParameterWhat to enterWatch out
Branch NameThe LZA release branch, for example release/v1.17.0Changing it later is how you upgrade
Enable Approval StageYes (default) and an approval email listSomeone must approve every run
Management / LogArchive / Audit Account EmailThree unique, monitored email addressesAn email address can belong to only one AWS account
Control Tower EnvironmentYes for the Universal Configuration—
Accelerator Resource name prefixLeave AWSAcceleratorChanging it after install causes stack failure
Configuration Repository LocationS3, CodeCommit, or CodeConnectionsCannot be changed after install
Enable Diagnostics PackYesHelps troubleshooting

Deployment day checklist

Common first-deployment failuresReused or mistyped email addresses; service quotas (for example CodeBuild concurrency) too low in new accounts; a home Region without Control Tower; editing an LZA-managed resource by hand; invalid Suricata syntax in firewall rules. The pipeline's CodeBuild logs tell you which stack and resource failed.
Universal Configuration · Module 5

The Universal Configuration

An AWS-published, ready-to-deploy set of LZA configuration files that turns an empty LZA into a well-architected, security-focused landing zone.

What it is

The LZA Universal Configuration (UC) lives on GitHub at aws/lza-universal-configuration. It is based on the AWS Well-Architected Framework, the AWS Security Reference Architecture and the "Organizing your AWS environment using multiple accounts" whitepaper. Instead of starting from a blank slate, a customer starts from proven patterns and adjusts.

Opinionated

Makes sensible decisions for you: account layout, guardrails, security services, logging, networking.

Built on Control Tower

Uses Control Tower as the foundation and layers extra controls on top.

Meant to be adjusted

A strong starting point. Customers should review every setting against their own requirements.

How it's packaged: modules + replacements

A UC release is assembled from building blocks. Each release ships ready-made zips for common combinations; under the hood the recipe is:

1. Base module

modules/base/default: accounts, OUs, guardrails, IAM, security, logging, backup, budgets. Always used.

2. Network module

Pick one: hub-and-spoke (default, a VPC per workload account) or shared-vpc (central VPCs shared by subnet).

3. Partition replacements

Adjust for where it runs: AWS standard (commercial), AWS GovCloud (US), or the AWS European Sovereign Cloud.

The partition files remove or change what isn't available there. For example, the GovCloud variant leaves out Amazon Macie, central root user management and one Config rule, because those aren't available in GovCloud (US).

What the UC deploys: a tour

AreaHighlightsConfig file
Accounts & OUs6 core accounts; Security, Infrastructure, Workloads (Sandbox/Dev/Test/Prod) and Suspended OUsaccounts, organization
Control TowerLanding zone version 4.0, Identity Center access, 11 extra Control Tower controls (detective checks)global
Guardrails8 SCPs (including quarantine for new accounts), 1 RCP for a data perimeter, 1 declarative policy blocking VPC internet access by default, tag and backup policiesorganization
IdentityIAM Identity Center delegated to the SharedServices account; standard Backup and EC2 SSM roles; a sample permission boundaryiam, global
Security servicesGuardDuty, Security Hub (AWS Foundational Security Best Practices, NIST 800-53 Rev 5, CIS v3.0.0), Macie, AWS Config rules, IAM Access Analyzer, EBS default encryption, S3 Block Public Access, all delegated to the Audit accountsecurity
LoggingControl Tower organization trail; all CloudWatch Logs replicated to the Log Archive account; VPC, Transit Gateway, firewall and DNS query logsglobal, network
NetworkingTransit Gateway hub, AWS Network Firewall inspection, central ingress/egress, central VPC endpoints, IPAM, DNS Firewall, 2 Availability Zonesnetwork
Backup & costTag-driven AWS Backup plans (continuous to monthly), backup vault per account, a $2,000 monthly budget alert, Cost and Usage Reportorganization, global

What a customer must change before deploying

Controls Test Pack (new in v1.4.0)

Each UC release now includes a Controls Test Pack (lza-universal-config-cats-<version>.zip) that runs on CATS, the Control Acceptance Testing Suite. After deployment it checks the live environment against all 193 controls in the LZA Compliance Workbook and reports results that can be mapped to a compliance framework in OSCAL format. It deploys temporary test resources and deliberately tries actions the SCPs and RCPs should block, so run it in a separate test landing zone first, or before real workloads move in.

RememberUC = base module + one network module + partition replacements. Update emails, Region, CIDRs and firewall rules, push, then prove it works with the Controls Test Pack.
Universal Configuration · Module 6

Accounts, OUs and guardrails

Who lives where, and which rules apply to them. Click any OU or account on the map.

Select an account or OU

Each one shows its purpose, what the UC deploys there, and which policies apply.

Layers of protection

The UC uses defense in depth: several kinds of controls that overlap, so one gap doesn't expose everything. Click a layer.

The SCPs in the Universal Configuration

PolicyAttached toIn plain English
How SCPs workAn SCP never grants permission; it sets the maximum. If an SCP denies an action, no IAM policy in that account can allow it. The UC's SCPs exempt LZA's own roles (names starting with the accelerator prefix and the Control Tower execution role) so the pipeline can still manage resources. The management account is never restricted by SCPs.
Limits to rememberUp to 10 SCPs and 5 RCPs can be attached directly to each OU or account. An SCP can be up to 10,240 characters; an RCP up to 5,120. LZA minifies policies before uploading them.
Universal Configuration · Module 7

Networking: hub and spoke

All traffic between VPCs, to the internet and to AWS services goes through one hub and gets inspected by a firewall. Play the traffic flows to see exactly how.

The pieces

Transit Gateway

The central router in the Network account. Every VPC attaches to it. Two route tables decide where traffic goes: spoke (everything → inspection) and firewall (back to the right VPC, or out to egress).

Inspection VPC

In the Network account. Holds AWS Network Firewall, which checks traffic against Suricata-style rules. The default rule set ends with drop everything.

Egress VPC

In the Perimeter account. Shared NAT gateways (one per AZ) and an internet gateway for outbound internet access.

Ingress VPC

In the Perimeter account. Public subnets for customer-deployed load balancers that receive traffic from the internet. LZA doesn't deploy the load balancers.

Endpoints VPC

In the Network account. Central interface endpoints (EC2, SSM, KMS, CloudWatch Logs, ECR and more) so workloads reach AWS APIs privately.

Workload VPCs

One VPC per account in the Dev, Test and Prod OUs, created automatically from VPC templates, with app, data and TGW subnets in two AZs.

Interactive network diagram

Pick a traffic flow to animate it, or click a VPC for details.

Pick a flow or a VPC

Orange dashed lines show the path. Numbers mark each hop.

The most common surpriseOut of the box, workloads cannot reach the internet. The firewall rules allow traffic to the central endpoints (DNS and HTTPS) and to Shared Services on port 443, block cross-environment traffic (Dev, Test and Prod can't talk to each other), and end with a default drop. To allow internet access, add pass rules (for example by domain name) to firewall-rules/rules.txt.

IP address plan (IPAM)

AWS IPAM hands out address ranges from a hierarchy of pools, so VPCs never overlap. These are the UC defaults from replacements-config.yaml. Click a pool.

Plan this earlyChanging CIDRs after deployment means replacing VPCs. Check the plan against on-premises and partner networks before day one, and leave room for more Regions.

Workload VPC layout (from the VPC template)

Every Dev, Test and Prod account gets a /22 VPC (1,024 addresses) from its environment's pool:

app-a /24app-b /24data-a /26data-b /26free

App subnets /24 (251 usable IPs each), data subnets /26 (59 usable each), TGW attachment subnets /28. AWS reserves 5 addresses in every subnet. Each environment pool is a /14, so it fits 256 of these VPCs (one per account).

Hub and spoke vs shared VPC

Hub and spoke (default)Shared VPC
Workload networkingEach workload account gets its own VPC from a templateThree central VPCs (shared-dev, shared-test, shared-prod) in the Network account; subnets shared to accounts with AWS RAM
Who controls routingAccount teams own their VPC (within guardrails)The central network team
IP efficiencyModerate: every account reserves a whole VPCHigh: accounts share address space
IsolationStrong account-level boundariesWorkloads in the same environment share a VPC
Choose whenDevOps teams, clear ownership, strong isolationTight IP space, central network operations, chatty workloads

More network services in the UC

  • DNS Firewall (Route 53 Resolver) on every VPC, blocking domains on AWS-managed threat lists (aggregate threats, malware, botnet command-and-control).
  • DNS query logging, VPC Flow Logs (all traffic, custom fields, 30 days in CloudWatch then centralized) and Transit Gateway Flow Logs.
  • Default VPCs deleted in every account and Region (they have open internet access).
  • VPC Block Public Access declarative policy: Security OU, Dev/Test/Prod OUs and the Network and SharedServices accounts can't use internet gateways. Perimeter is excluded because ingress and egress need them.
  • Two Availability Zones (a and b) for every VPC. Adding a third AZ is a supported customization (more subnets, NAT gateway and firewall endpoint).
  • Hybrid connectivity (Direct Connect or VPN to on-premises) is documented but not deployed by default; it attaches to the same Transit Gateway.
  • Optional Transit Gateway encryption support (LZA v1.17.0+, commented out by default) for encryption in transit with VPC Encryption Controls.
RememberSpoke route table sends everything to inspection; the firewall route table sends it on to the destination VPC or to egress. Default firewall action: drop. Internet out = Egress VPC (Perimeter). Internet in = Ingress VPC (Perimeter). AWS APIs = Endpoints VPC.
Universal Configuration · Module 8

Security, identity, logging and backup

The services that watch, record and protect the environment, and which account runs each one.

Who does what: delegated administration

Instead of running everything from the powerful management account, the UC delegates each service to a specialist account. That keeps the management account quiet and limits who needs access to it.

AccountAdministersWhy
Audit (Security OU)GuardDuty, Security Hub, Macie, AWS Config aggregation, IAM Access AnalyzerThe security team sees findings for every account in one place
Log Archive (Security OU)Central log buckets (CloudTrail, CloudWatch Logs, access logs, ELB logs)Logs are stored where workload owners can't change or delete them
SharedServices (Infrastructure OU)IAM Identity CenterSeparates identity administration from organization management
Network (Infrastructure OU)Transit Gateway, IPAM, Network Firewall, endpointsNetwork team owns shared connectivity

Security services turned on

How logs get to the Log Archive account

Click each step of the pipeline.

Identity: how people sign in

  1. A user signs in with the customer's identity provider (for example Microsoft Entra ID, Okta or Active Directory), federated to IAM Identity Center with SAML 2.0 and SCIM.
  2. Identity Center (administered from the SharedServices account) shows the AWS access portal with the accounts and permission sets the user's groups are assigned.
  3. The user picks an account and role and gets temporary credentials. No long-lived IAM users: an SCP blocks creating them.
Also deployedA strict IAM password policy (14+ characters, 24 remembered, 90-day expiry) for any remaining IAM users, a Backup-Role and an EC2-Default-SSM-Role in every account, and a sample permission boundary (End-User-Policy) teams can use to limit roles they create.

Backups and tags

Backups are driven by a tag. Put BackupPlan on a resource and AWS Backup picks it up automatically in Infrastructure and Workloads accounts:

BackupPlan tag valueScheduleRetention
Continuous (S3 only)Continuous, point-in-time35 days
HourlyEvery hour at :05365 days (30 in standard storage, then cold)
Daily05:00 daily365 days (30 in standard, then cold)
WeeklySundays 05:00365 days (30 in standard, then cold)
Monthly1st of the month 05:00730 days (60 in standard, then cold)

Tag policies make sure the tag value is valid; they don't force every resource to have it. Resources without the tag aren't backed up.

Cost visibility

  • A monthly AWS Budget of $2,000 with email alerts at 50%, 75%, 90% and 100%.
  • A Cost and Usage Report in Parquet format for detailed analysis.
  • Security notifications in three SNS topics (High, Medium, Low) so the right people get the right alerts.
RememberAudit = security findings. Log Archive = logs. SharedServices = Identity Center. Network = connectivity. Logs flow CloudWatch → Kinesis → Firehose → S3 in Log Archive. Backups follow the BackupPlan tag.
Customer ready · Module 9

Day 2: running the landing zone

Deployment is day one. These are the tasks the customer's team will do every week, and the rules that keep the environment healthy.

The golden rule

Change the config, not the consoleEverything LZA manages should be changed in the configuration files and deployed through the pipeline. Manual changes are either blocked by SCPs, reverted on the next run, or cause the next run to fail. Workload resources (EC2, databases, applications) are managed normally by the workload teams.

Common tasks

Add a new workload account
  1. Reserve a unique email address.
  2. Add an entry under workloadAccounts in accounts-config.yaml with a name, email and OU (for example Workloads/Dev).
  3. Push and approve. The Prepare stage creates it through Control Tower Account Factory; quarantine SCP blocks it until the pipeline finishes; the VPC template for its OU gives it a VPC automatically.
Add a new OU

Add it to organizationalUnits in organization-config.yaml, then reference it in policies, VPC templates and share targets as needed. Manage OUs only through LZA after go-live.

Allow workloads to reach an internet site

Add a pass rule above the default drop in firewall-rules/rules.txt, for example a TLS rule that matches an SNI domain like .github.com. Invalid Suricata syntax fails the pipeline, and the firewall is shared by every environment, so try new rules in a separate test landing zone first if you have one.

Change or add an SCP

Edit or add the JSON file in service-control-policies/ and reference it in organization-config.yaml with its deployment targets. Keep under 10,240 characters and remember the 10-SCPs-per-target limit. The UC turns on scpRevertChangesConfig, so manual SCP edits in the console are reverted.

Upgrade LZA to a new version
  1. Read the release notes for every version between yours and the target.
  2. Run the Core pipeline once on your current version and fix any failures (a clean baseline).
  3. Update the installer CloudFormation stack: replace the template with the latest one and set Branch Name to the new release branch.
  4. The stack updates in about 10 minutes; the Installer and Core pipelines then run (another 20–60 minutes, depending on accounts and Regions).
Upgrade the Universal Configuration

UC releases come with a changelog and upgrade notes for each change. Compare the new release with your current config (a diff tool helps), adopt the changes you want, and push. The UC is configuration, not software: nothing updates until you merge and deploy.

Decommission an account

Move it to the Suspended OU in the config. The Suspended guardrail SCP denies everything, so nothing in it can run while you close it.

Troubleshooting a failed pipeline

  1. Open the Core pipeline in CodePipeline and find the failed stage.
  2. Open the CodeBuild logs for that action: they name the account, Region and CloudFormation stack.
  3. Open that stack's Events in the target account to see the exact resource and error.
  4. Fix the cause (usually the config, a quota, or a manual change) and choose Retry or release a new change.
Diagnostics packIf enabled at install, the diagnostics pack can collect information to share with AWS Support or the LZA GitHub issues.

Day 2 do's and don'ts

Do

  • Keep the config in version control with pull-request reviews
  • Keep the approval stage on: each run changes every account, Dev and Prod together
  • Upgrade regularly; read release notes
  • Try config changes and upgrades in a separate test landing zone (its own organization) first
  • Watch Security Hub findings and budget alerts

Don't

  • Edit LZA-created resources in the console
  • Change the accelerator prefix or config repository location after install
  • Create OUs or accounts outside LZA
  • Delete the default-deny firewall rule to "fix" connectivity
  • Skip versions without reading every release's notes
Customer ready · Module 10

Briefing a customer

How to explain LZA and the Universal Configuration to different audiences, collect the inputs you need, and answer the questions you will get.

The 30-second pitch

"The Landing Zone Accelerator is a free AWS solution that builds a secure, multi-account AWS environment from configuration files, and keeps it that way through an automated pipeline. With the Universal Configuration, you start from AWS's best-practice design: separate accounts for security, networking and each environment, guardrails that prevent risky actions, centralized logging and threat detection, and inspected networking. You get a well-architected foundation in days instead of months, and every change after that is reviewed and repeatable."

Tailor the message

Suggested briefing agenda (60 minutes)

MinTopicUse from this guide
5Their goals and constraints (compliance, timeline, existing AWS use)Discovery questions below
10Why multi-account; what a landing zone isModule 1 analogy
10How LZA works: config files and pipelinesModule 3 big picture
15What the Universal Configuration builds: accounts, guardrails, network, securityOrg map and network diagram
10Deployment plan, timeline and responsibilitiesModule 4 phases
10Questions, decisions needed, next stepsInputs checklist

Discovery questions

  • Which compliance frameworks matter (for example NIST 800-53, CIS, FedRAMP, DoD SRG, HIPAA)?
  • Do you already use AWS? Is there an existing organization, Control Tower, or accounts to bring in?
  • Which Regions (and partitions, such as GovCloud) do you need now and later?
  • What identity provider will people sign in with?
  • What IP ranges are used on-premises and in other clouds? Will you connect with Direct Connect or VPN?
  • Which internet destinations do workloads need (package repositories, SaaS APIs)?
  • Who will own the landing zone day to day: a platform team, network team, security team?
  • How do you manage change today (pull requests, change advisory board)?

Inputs to collect before deployment

Questions you will get

Cost: what drives the bill

LZA itself is free. The AWS services it enables are not. The biggest always-on items in the UC network are hourly charges. Rough us-east-1 list prices to frame the conversation (always build a real estimate with the AWS Pricing Calculator):

Item in the UCQuantity (2 AZs)Approximate list price
AWS Network Firewall endpoints2 (one per AZ)$0.395 per endpoint-hour ≈ $570/month, plus $0.065/GB processed
NAT gateways (egress)2Hourly plus per-GB charges
Transit Gateway attachmentsOne per VPC (5 core + each workload VPC)$0.05 per attachment-hour ≈ $36/month each, plus per-GB
Interface VPC endpoints13 services × 2 AZsHourly per endpoint per AZ, plus per-GB
Security servicesEvery accountUsage-based: GuardDuty, Security Hub, Config, Macie scale with resources and data
Logging and storageCentral buckets, Kinesis, FirehoseScales with log volume

Centralizing NAT, firewall and endpoints is cheaper than putting them in every VPC: that is one of the design goals of the hub. The included $2,000 budget alert helps spot surprises early.

Shared responsibility for the landing zone

AWS provides

  • The LZA solution, the Universal Configuration and their documentation
  • The underlying services and their security
  • Fixes and new versions through GitHub releases

The customer owns

  • Reviewing and approving every setting
  • Running, upgrading and monitoring the landing zone
  • Workload security, data classification and their compliance authorization (for example an ATO)
Practice

Flash cards

Click a card (or press Space) to flip it. Mark whether you knew it; cards you miss come back. Arrow keys move between cards.

Practice

Readiness quiz

26 questions covering the whole guide. Each answer explains why. Aim for 85% or better before briefing a customer.

0 / 0 answered
Reference

Glossary

Every acronym and term in one place.

Reference

Sources and further reading

Official sources used for this guide. Check them for the latest details before a deployment.

Universal Configuration details (accounts, policies, CIDRs, firewall rules, endpoints, backup plans) were read directly from the v1.4.0 configuration files and documentation.