Every cloud provider fails eventually. On 12 and 13 November 2026, Microsoft 365 email was down for about 24 hours worldwide, and Microsoft’s own safeguards between regions failed to contain it. Our live coverage of the Microsoft 365 outage has the full timeline. Companies with a business continuity plan for cloud systems kept invoicing, selling and supporting customers. Companies without one waited.
This guide shows what your cloud SLA covers, what stays your responsibility, and how to build a cloud outage plan in seven steps.
What Your Cloud SLA Actually Covers
A service level agreement promises uptime and pays service credits when the provider misses it. It does not keep your business running.
Provider | Uptime commitment | Downtime allowed in a 30-day month |
Microsoft 365 | 99.9% per service, per month | About 43 minutes |
NetSuite | 99.7% outside scheduled maintenance | About 2 hours 10 minutes |
Credits refund part of a monthly fee. They do not recover from a lost day of orders, a missed payroll run or a customer who went elsewhere.
What Stays Your Responsibility
Microsoft’s shared responsibility model is clear: whatever the cloud service, you keep responsibility for your data, your user accounts, access management and the devices that connect to the service. The provider runs the platform. Keeping your business processes running is yours.
How to Build a Business Continuity Plan for Cloud Systems
1. List Your Critical Processes
Start with processes, not systems: taking orders, invoicing, paying suppliers, running payroll, and supporting customers. Rank them by how fast a stoppage costs money.
2. Set Recovery Targets
Give each process two numbers. The recovery time objective (RTO) is how long the process can be down before the business is harmed. The recovery point objective (RPO) is the point in time your data must be recovered. Both terms come from NIST’s contingency planning guide, SP 800-34.
3. Map Every Dependency
List every system each process touches: email, identity and sign-in, ERP, payments, banking, domain and DNS, and integrations. The November outage showed how far one dependency reaches. When Exchange Online failed, parts of Teams, SharePoint, Copilot, Defender and Purview failed with it.
4. Define a Fallback for Each Dependency
Dependency | Fallback |
A second channel for staff and customers: phone tree, SMS and a public status page | |
Sign-in | Documented emergency administrator accounts that do not rely on single sign-on |
ERP approvals | Approval queues inside the ERP, not only in email |
Files | Offline copies of the documents needed to keep trading |
Integrations | Failure alerts outside email and a manual workaround for each flow |
5. Keep Backups You Control
Back up Microsoft 365 mail and files with a separate backup service, and schedule regular exports of ERP master data and open transactions. For systems you run yourself, define infrastructure as code so you can rebuild in another region. Our Azure DevOps services and AWS DevOps services teams set this up and test the rebuild.
6. Write the Communication Plan
Decide who declares an incident, where staff get updates when email is down, and who speaks to customers and suppliers. Pre-write those messages now, while nobody is under pressure.
7. Test the Plan Twice a Year
Run a tabletop exercise: pick a scenario such as “email is down for 24 hours” and walk through each critical process. Fix what breaks on paper before it breaks for real.
IT Disaster Recovery vs Business Continuity
IT disaster recovery restores systems. Business continuity keeps the business operating while systems are down. For software you rent, the provider restores the service and you own continuity. For systems you run, you own both, and our DevOps team builds the recovery side into your infrastructure from the start.
Lessons From the Microsoft 365 Outage
Provider safeguards fail too: Microsoft’s regional boundaries did not contain this outage, so the impact became global.
Recovery takes longer than the fault: Microsoft found the cause overnight, yet servers kept failing as queued traffic returned, and full recovery took most of the next day.
Email is a dependency for almost everything: Approvals, alerts, customer replies and reports all stopped with it.
Conclusion
A cloud SLA pays credits, but a business continuity plan keeps you trading. List your critical processes, set recovery targets, map dependencies, define fallbacks and test them. Talk to our team and we will help you build and test a cloud outage plan around the systems your business runs on.
