Continuity in Microsoft 365 and Cloud Infrastructure

Continuity in Microsoft 365: shared responsibility, the limits of the recycle bin and retention policies, MFA, emergency access accounts and hybrid identity sync.
Technical summary
Technology
Microsoft 365, Microsoft Entra ID
Platform
Cloud and hybrid infrastructure
Topic
Business continuity and identity security
Level
Start
In this article8 sections

Businesses that move to Microsoft 365 often assume: “Everything is in the cloud now; backup and continuity are Microsoft’s job.”

That is partly true. Microsoft is responsible for service availability, data centers and the platform infrastructure. The data itself, user identities, access rights and the configuration of the Microsoft 365 environment (tenant), however, are largely the business’s responsibility.

The real continuity question is this: after a user mistake, a malicious mass deletion or a compromised admin account, how quickly and completely does the business recover? The answer lies not in the platform but in your configuration.

The shared responsibility model

In cloud services, responsibility is split between the provider and the customer. In the SaaS model, the physical infrastructure, network and application platform sit with the provider. Information and data, devices, accounts and identities stay with the customer in every service model.

In practice this means: Microsoft keeps Exchange Online and SharePoint running. Noticing an accidentally deleted folder in time and bringing it back, deciding which data is kept for how long and protecting accounts is your job.

A common gap is that this split has never been written into the contract or internal procedures. When a loss occurs, nobody owns the question “who will bring it back?”

The recycle bin is not a backup

Deleted items in Microsoft 365 don’t disappear immediately; they are kept recoverable for a certain period. In Exchange Online, deleted items are kept for 14 days by default, and this can be extended to a maximum of 30 days per mailbox. In SharePoint and OneDrive, deleted content stays in the recycle bins for a limited time.

These mechanisms help with everyday user mistakes; they don’t replace a backup strategy:

  • If the loss is noticed after the retention period has expired, a restore may no longer be possible.
  • Local files encrypted by ransomware can also affect their cloud copies through OneDrive sync.
  • Deliberate bulk deletions made with a privileged account can target the recycle bins too.
  • The recycle bin doesn’t offer a consistent copy as of a specific date.

OneDrive’s file restore feature helps you go back to a specific date, but it works per user and still depends on retention periods. In an incident affecting hundreds of users, restoring one by one exceeds the RTO target.

Retention policies versus backup

Microsoft Purview retention policies keep content from being deleted for a set period and are needed for regulatory and compliance requirements. But their purpose is compliance; they are not a backup solution designed for fast, bulk restores. Retained content can be searched and exported, but it doesn’t return a mailbox or site structure to its previous state in a single step.

For critical data, a separate backup solution that allows restoring data as of a specific point in time should be considered in addition to retention policies. That solution’s admin account should also be kept separate from Microsoft 365 admin accounts; it must not be possible to delete both the data and the backup with the same compromised account.

Identity security is part of continuity

In the cloud, identity is the new security perimeter. A single compromised admin account means access to mailboxes, files and the entire environment configuration. You are not behind a firewall. The door is the sign-in screen.

Microsoft’s own guidance is that multi-factor authentication (MFA) and blocking legacy authentication protocols stop the vast majority of common identity attacks. The basic steps:

  • Require MFA for all users. Make no exceptions, especially for admin accounts.
  • Block legacy authentication. Protocols that don’t support MFA are an open door for attackers.
  • Use security defaults or Conditional Access. Security defaults provide basic protection without extra licensing; for more detailed rules, Conditional Access, included with Microsoft Entra ID P1, can be used.
  • Separate admin accounts. The account used for daily work and the admin account should not be the same.

A common configuration mistake: Conditional Access rules are created but left in “report-only” mode and never enabled. The rule list looks full, but there is no protection. Watch the rules in report mode for a few days, see their effect in the sign-in logs, then turn them on.

Emergency access accounts

A problem with authentication methods or a misconfigured access rule can lock every administrator out of the environment. That is why Microsoft recommends creating two or more emergency access (break-glass) accounts.

These accounts should be cloud-only, protected with strong, phishing-resistant methods, monitored and never used for daily work. One more thing: set up monitoring that alerts administrators whenever these accounts sign in, and test at regular intervals that the account really works.

Access management

Continuity is not just about bringing data back; the right people also need to keep access to the right data.

  • Permissions should follow the principle of least privilege, and admin roles should be reviewed regularly.
  • Files shared externally and guest users should be checked periodically.
  • For employees who leave, account closure and the handover of mailbox and OneDrive data should follow a defined process.

A point to watch with leavers: when the user is deleted, their OneDrive data is also deleted at the end of the retention period. Do the handover before deleting the account.

Things to watch in hybrid environments

In many businesses, users are kept in on-premises Active Directory and synchronized with Microsoft Entra ID. In this setup, cloud access depends on the health of the on-premises infrastructure.

When the sync server fails, password changes or new users may not reach the cloud. This server should be treated as a critical system, backed up and monitored. Keeping a second Microsoft Entra Connect server in staging mode is a well-known way to speed up the switchover if the primary fails. The health of the domain controllers is also part of cloud continuity: an AD with broken replication means wrong identity data in the cloud too. Decide which work must continue during an outage and the acceptable waiting time in a business impact analysis and business continuity plan.

Preparing by scenario

Scenario Possible impact Preparation
A user accidentally deletes a folder A single user or team is affected Recycle bin and restore-from-backup procedure
Ransomware encrypts OneDrive files All synced files may be affected Point-in-time backup, endpoint protection
An admin account is compromised The entire environment is at risk MFA, separate admin accounts, sign-in monitoring
The identity sync server fails User changes don’t reach the cloud Server backup and documented rebuild steps
An access rule is misconfigured Administrators cannot access the environment Emergency access accounts

Every row in the table should have an owner and a written list of steps. During an incident, you don’t look for the procedure; you follow it.

Short checklist

  1. Has a backup approach been defined for critical Microsoft 365 data?
  2. Have retention periods been set according to business and regulatory needs?
  3. Is MFA required for all users?
  4. Have legacy authentication protocols been blocked?
  5. Are there at least two emergency access accounts, and are they monitored?
  6. Are admin roles and guest access reviewed regularly?
  7. In a hybrid setup, is the sync server backed up and monitored?
  8. Have restore scenarios been tested at least once?

The platform working is not enough

Microsoft 365 is a highly available platform. But the platform running does not mean the business’s data and identities are protected.

Real continuity needs three things together: recoverable data, protected identities and access under control. If one is missing, the plan is incomplete.

The Eretek Bilişim Hizmetleri team plans backup, identity security and access management in Microsoft 365 and hybrid infrastructures around the business’s continuity goals.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek System Support

Server installation, configuration, incident response and scheduled maintenance. Active Directory, virtualization and backup are managed with the same discipline.

Knowledge Center

Other articles