Cloud Backup High Availability via Shared Object Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional applications face challenges in achieving high availability in public cloud environments without requiring significant architectural changes, especially for mission-critical applications like Data Domain Virtual Edition, due to limitations in shared disk offerings and data consistency across multiple instances.
Innovation Solution
A high availability system is implemented using shared disks, object storage bucket consistency features, and existing high availability features like Dell's Data Domain Restorer, which creates active and passive nodes in the public cloud, leveraging cloud technologies such as Amazon Web Services, Azure, and Google Cloud Platform to ensure seamless failover and data consistency without architectural changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional applications are deployed in public cloud environments, then cloud computing resources and scalability are utilized, but high availability cannot be achieved without significant architectural changes
Solution Approach 1:
The system segments the backup/restore application into multiple independent instances running on separate virtual machines. Each instance can operate autonomously, allowing one instance to take over if another fails. This segmentation enables high availability without requiring fundamental architectural changes to the application itself.
Solution Approach 2:
The patent creates duplicate copies of the backup/restore application on multiple virtual machines. These copies maintain identical functionality and can interchange roles as active or passive nodes. By copying the application rather than redesigning it, the system achieves high availability while minimizing architectural modifications.
2Adaptability or versatility
If shared disk offerings are used for multiple instances, then data sharing between nodes is enabled, but data consistency across multiple instances becomes problematic
Solution Approach 1:
The patent introduces a shared object storage bucket as an intermediary between multiple application instances. Instead of allowing direct access to shared disks which causes consistency issues, the object storage bucket mediates data access by providing a standardized interface that ensures data consistency while enabling sharing between nodes.
Solution Approach 2:
The system uses a universal object storage service that can be accessed by multiple virtual machines and application instances simultaneously. This universal storage layer provides both data sharing capability and consistency guarantees, resolving the conflict between adaptability and reliability.
3Reliability
If active and passive nodes are implemented for failover, then high availability is achieved, but storage costs increase
Solution Approach 1:
The patent merges the storage resources of active and passive nodes by having both instances access the same shared object storage bucket. Instead of duplicating storage capacity across nodes, the system combines storage access through a shared bucket, reducing total storage requirements while maintaining failover capability.
Solution Approach 2:
Rather than copying entire data sets to multiple nodes, the system copies only the application instances themselves. The actual data resides in a single shared object storage bucket that all instances access. This approach minimizes storage duplication while enabling active-passive failover.
Data Source
AI summary
Providing high availability for backup and restore applications in public clouds is described. A system creates a first virtual machine and a second virtual machine for at least one zone of a public cloud. The system creates a disk in the public cloud, and enables a selection to attach the disk as a shared disk to the first virtual machine and/or the second virtual machine; The system creates object storage that stores backup data for the first virtual machine and/or the second virtual machine. The system deploys the first virtual machine as an active node that executes a backup and restore application and the second virtual machine as a passive node that executes the backup and restore application when the active node fails to execute the backup and restore application.