ATC323 Architecting Highly Available, Geo-Redundant Cloud Solutions with Windows Azure Graham Elliott Asia Time Zone Director Global Technical Evangelism and Development Team Microsoft

Why? Stuff Happens You are ultimately responsible for: Fault Tolerance Disaster Recovery High Availability What do you want to protect against?

Recovery Point Objective (RPO) Disaster Last Backup or Point when data can be consistently rolled back RPO How far back?

Recovery Time Objective (RTO) Disaster Last Backup or Point when data can be consistently rolled back RPO How far back? Systems fully recovered RTO How much time?

Other Considerations RPO RTO Performance Consistency Geographic location Cost SLAs Compliance Last Backup or Point when data can be consistently rolled back Systems fully recovered Disaster How far back? How much time? RPO RTO

High Availability and Disaster Recovery Hot HA (Live/Live Config) No Downtime Hot DR (Least Common) > 5 Minutes > 1 Hour Warm DR Local Fault Tolerance with Cold DR > Few Hours $ $$ $$$ $$$$

Windows Azure Geo-DR Architectures

Cold DR (with local fault tolerance) Client Data Center 1 Redirect DNS Data Center 2 Load Balancer Load Balancer Cloud Service … Cloud Service Cloud Service … Cloud Service Fault Domain Fault Domain Deployment Package Fault Domain Fault Domain Redeploy SQL Database Azure Storage Azure Storage SQL Database Geo-Replication* Restore BACPAC Custom / 3rd Party Backup BACPAC from DB Copy BACPAC from DB Copy

Demo SQL Database Export Storage Geo-Replication

Regions, Sub Regions and Data Centers North America Region Europe Region Asia Pacific Area N. Europe Sub-Region N. Central – U.S. Sub-Region NE. China (via 21Vianet) Sub-Region West – U.S. Sub-Region East – U.S. Sub-Region East Japan Sub-Region W. Europe Sub-Region West Japan Sub-Region E. China (via 21Vianet) Sub-Region S. Central – U.S. Sub-Region E. Asia Sub-Region S.E. Asia Sub-Region Major datacenter CDN node Live sub-region Announced sub-region Partner-operated sub-region East Australia Sub-Region Southeast Australia Sub-Region

Other Building Block Business Continuity TechEd 2013 5/12/2019 10:09 AM Other Building Block Business Continuity virtual machines Availability / load-balanced sets. No charge if stopped. SQL: Most HA/DR support in some form. DR = Recreate from script/image/disk. identity HA and resilience within a single data center ACS 1.0 (deprecated): Not recoverable. ACS 2.0: Offsite daily backups provided. Powershell based export/import. web sites Built in FT for Shared and Standard modes. Failure detection and FT with single instance. Automatic load balancing and autoscale. DR = Redeploy. HA and resilience within a single data center. DB backend as per SQL Database. DR = Redeploy. mobile caching Caching: HA across fault domains. Shared Caching: HA not supported. DR = Redeploy. media HA and resilience within a single data center. DR should focus on data stores. DR = Redeploy. service bus Service namespaces cannot span DCs. No Auto geo-replication. Relayed: geo-replication of endpoints sample. Brokered: Active & passive replication sample. Client: Durable client side queue. SQL Reporting HA and resilience within a single data center. DR should focus on data stores. DR = Redeploy. © 2013 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.

Warm DR Client … … Geo-Replication* Restore BACPAC Custom / 3rd Party Data Center 1 Traffic Manager Data Center 2 Load Balancer Load Balancer Cloud Service … Cloud Service Cloud Service … Cloud Service Fault Domain Fault Domain Fault Domain Fault Domain SQL Database Azure Storage Azure Storage SQL Database Geo-Replication* Restore BACPAC Custom / 3rd Party Backup BACPAC from DB Copy BACPAC from DB Copy SQL Data Sync

Demo Traffic Manager SQL Data Sync

Hot DR – Azure SQL Database Client Data Center 1 Traffic Manager Data Center 2 Load Balancer Load Balancer Cloud Service … Cloud Service Cloud Service … Cloud Service Fault Domain Fault Domain Fault Domain Fault Domain SQL Database Azure Storage Azure Storage SQL Database Custom Replication or HA Proxy Service Bus Service Bus

Hot DR – SQL Server in a VM Client Data Center 1 Traffic Manager Data Center 2 Load Balancer Load Balancer Cloud Service … Cloud Service Cloud Service … Cloud Service Fault Domain Fault Domain Fault Domain Fault Domain SQL Server in a VM Azure Storage Azure Storage SQL Server in a VM Custom Replication or HA Proxy Backup to URL / Restore or Database Mirroring

SQL Server in Azure HA/DR - Scenarios HA within Azure (single Data Center) DR across Azure Data Centers DR between On-Premise and Azure

SQL Server HA within Azure Why? Failure detection for SQL Server not just VM Azure service healing Azure upgrades S P S P VM VM Windows Azure (Single Data Center)

Technologies Supported Technology SQL Version Zero Data Loss Scope # Secondaries Automatic Failover Readable Secondaries Availability Groups SQL Server 2012 Yes (Sync Mode) DB(s) 4 (w/ additional cluster member) Database Mirroring SQL Server 2008 R2 (High Safety Mode) DB 1 (w/ Witness) Limited (database snapshots)

SQL Server DR across Data Centers P Why? If you use multiple disks for SQL Server If Azure storage DR doesn’t meet your needs Freedom to test / execute DR Windows Azure Windows Azure

Technologies Supported Technology SQL Version Zero Data Loss Scope # Secondaries Automatic Failover Readable Secondaries Database Mirroring SQL Server 2008 R2 SQL Server 2012 Yes (High Safety Mode) DB 1 (w/ Witness) No Backup to Url SQL Server 2012 SP1 CU2 N/A * Most customers use Async Mode (High Performance) to avoid impacting primary performance **Async Mode only supports Force Failover to make DR (and potential data loss) a conscious decision

Thanks !! Large Scale failures in the cloud are rare, but happen. Application Owners are ultimately responsible for availability and recoverability Balance cost and complexity of HA against the risks you are willing to bear. You also have the option to degrade features, and performance to add resiliency.

Resources Learning TechNet Developer Network 5/12/2019 10:09 AM Sessions on Demand http://channel9.msdn.com/Events/TechEd/Australia/2013 Virtual Academy http://www.microsoftvirtualacademy.com/ TechNet Developer Network Resources for IT Professionals http://technet.microsoft.com/en-au/ Resources for Developers http://msdn.microsoft.com/en-au/ © 2013 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.

5/12/2019 10:09 AM © 2013 Microsoft Corporation. All rights reserved. Microsoft, Windows and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION. © 2013 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.