Other Articles

Production Access Control and Management

Support > Tech Best Practices

Overview

This article explains how to control and manage access to your production environment. Production access is one of the most sensitive areas of infrastructure security because anyone who can access production may affect your live application and your customers' data.

This guidance applies to all cloud platforms. The examples use AWS, Microsoft Azure, and Google Cloud, but the underlying principles are the same across providers.

What Is Production Access Control?

Production access control is the set of rules and tools that determine who can access your live environment, how they access it, and what they can do once they are there. It covers the cloud console, command-line tools, running containers, and servers.

The goal is simple: only authorized people should be able to access production through approved methods, and every access event should be traceable.

Why Control Production Access?

  • Production contains live customer data and runs the services your customers depend on.
  • A mistake or a compromised account with production access can cause an outage or a data breach.
  • Unauthorized changes are difficult to detect and reverse when many people have broad access.
  • Shared or long-lived credentials are common targets for attackers and are difficult to trace to an individual.
  • Frameworks such as ISO 27001 and SOC 2 expect production access to be restricted and reviewed.

Benefits

  • Production resources are protected from accidental and unauthorized changes.
  • Every access event can be traced to a named individual.
  • Credentials are short-lived, reducing the potential damage if they are leaked.
  • When someone leaves the team, their access can be revoked quickly through a central identity system.
  • Troubleshooting remains possible through approved and logged methods.

How It Works

Strong production access control combines the following six controls.

1. Limit Access to a Small, Named Group

  • Grant production access only to people who need it to operate the system, such as members of the DevOps team.
  • Assign the highest-level cloud account ownership to a senior person, and do not use that account for routine work.
  • Maintain a written list of everyone with production access and document why each person needs it.
  • Do not grant developers or testers production access by default.

2. Sign In to the Console Through Single Sign-On

  • Use your company's central identity provider so that people sign in with their company identities.
  • Enforce multi-factor authentication (MFA).
  • Assign permissions according to each person's role, and avoid creating separate cloud users with individual passwords when centralized identity is available.

3. Use Temporary Credentials for the Command Line

  • Allow people to sign in through single sign-on and obtain short-lived credentials that expire automatically.
  • Avoid long-term access keys wherever possible.
  • Use role-based permissions to determine which commands each person can run.

4. Control Access to Running Containers

  • Use an approved tool to open an interactive session in a running container, avoiding the need to log in to the underlying server.
  • Require appropriate permissions for every session, and restrict session access to authorized personnel.
  • Record sessions and use them only for approved troubleshooting and maintenance activities.

5. Secure Server Access

  • Give each person their own SSH key and install it only on the servers they are authorized to access. Never share keys.
  • Restrict administrator accounts on servers to a very small, named group.
  • Keep management ports closed to all IP addresses by default.
  • Allow access from a person's current IP address only when needed, and remove the allowance afterward. An approved process or script can automate this.
  • Using a non-standard port may provide an additional layer of obscurity, but it must never be the only security measure.
  • Where possible, use a managed session service so that no inbound port needs to be open.

6. Log, Monitor, and Review

  • Record who accessed production, when the access occurred, and what actions were performed.
  • Review the list of people with access regularly, for example, every quarter.
  • Revoke access as soon as someone leaves the team or no longer needs it.

Applying These Controls in Your Cloud

The six controls apply across cloud platforms, but the services have different names. The examples below show where to start. Features vary by plan and version, so consult your provider's documentation.

  • AWS: Use AWS IAM Identity Center for console and command-line sign-in with temporary credentials, Amazon ECS Exec and AWS Systems Manager Session Manager for container and server sessions, and AWS CloudTrail for audit logs.
  • Microsoft Azure: Use Microsoft Entra ID for sign-in and Privileged Identity Management for time-limited access, Azure Bastion for server sessions, and the Azure activity log for audit records.
  • Google Cloud: Use Cloud Identity or Google Workspace for sign-in, OS Login and Identity-Aware Proxy for server access, and Cloud Audit Logs for audit records.
  • Other platforms: Look for equivalent features, such as single sign-on, short-lived credentials, audited session tools, and activity logs.

Common Use Cases

  • Setting up production access for a new application.
  • Reducing the number of people who can access production after a team change.
  • Replacing shared or long-lived credentials with individual, temporary credentials.
  • Allowing engineers to troubleshoot live issues safely.
  • Revoking access quickly when someone leaves the company.

Example Scenario

  • A small DevOps team is the only group with access to the production environment, while a senior leader holds the highest-level owner access to the account.
  • Team members access the cloud console through the company's sign-in portal and use multi-factor authentication.
  • For command-line tasks, they sign in through single sign-on and receive temporary credentials that expire automatically.
  • To investigate an issue in a running container, an authorized engineer opens an approved, logged session instead of logging in to the underlying server.
  • Servers are accessible only over SSH using each person's own key, and the SSH port is closed to everyone by default.
  • Before connecting, the engineer runs an approved script that temporarily allows access from their current IP address.
  • Every session is logged, and the list of people with access is reviewed regularly.

Best Practices

  • Maintain a written list of everyone with production access and review it regularly.
  • Grant production access to as few people as possible.
  • Protect cloud account owner credentials with multi-factor authentication, and do not use them for routine work.
  • Use single sign-on with multi-factor authentication for all human access.
  • Use temporary credentials instead of long-term access keys.
  • Give each person their own SSH key, and never share keys or logins.
  • Keep management ports closed by default, and allow access only from approved IP addresses for as long as necessary.
  • Use managed session tools where possible so that no inbound port needs to be open.
  • Log production sessions and review the logs regularly.
  • Revoke access as soon as it is no longer needed.

Common Mistakes to Avoid

  • Granting many people administrator access to production for convenience.
  • Using long-term access keys that are never rotated or revoked.
  • Sharing an SSH key or login among several people.
  • Leaving SSH or console management ports open to the entire internet.
  • Relying on a non-standard port as the only security measure.
  • Leaving an IP address allowlisted after the work is complete.
  • Using cloud account owner credentials for routine work.
  • Failing to revoke access when someone leaves the team or changes roles.
  • Failing to keep records of who accessed production.

Updated on 28 October, 2026