Have Your Cake and Eat It Too - Object-Level Recovery
In the last few weeks, Kenny covered how we’ve designed Rubrik to eliminate the effects of VMware application stun and JB covered why we built our own VSS Provider to take application-consistent snapshots. In this post, I’ll wrap up the app consistency series by discussing our object-level recovery and search capabilities. Maintaining application and data consistency in data protection requires a “three-layer cake” design. The base layer is our Converged Data Management platform, the next layer is the application tier, and the frosting and sprinkles are the objects/messages/files that are embedded inside the application and database.
Maintaining application and data consistency in data protection requires a “three-layer cake” design. The base layer is our Converged Data Management platform, the next layer is the application tier, and the frosting and sprinkles are the objects/messages/files that are embedded inside the application and database.
Our platform features a Distributed Task Framework, an engine that assigns and executes data across the cluster in a fault tolerant and efficient manner. It ensures all tasks are load balanced across the cluster and distributed to the nodes that contain the relevant data. This allows us to handle the ingestion of data across multiple nodes in parallel. Adding more nodes to the system results in a linear increase of data ingestion – more nodes can pick up the task of ingesting data from your data sources to be protected. On top of that, we utilize flash for an extra performance kick to maximize ingest rates to near-line rates.
In the middle is our own Microsoft VSS Provider, which provides rightly timed application-consistent snapshots. This is a highly orchestrated dance that facilitates the quiescing of the SQL or Exchange application ahead of the VMware file system snapshot to guarantee an application consistent restore point and can be used to truncate Exchange logs after job completion. Building our own VSS Provider vs. relying on the built-in VSS Provider offered by VMware allowed us to facilitate tighter application integration and leverage this when we start supporting SQL applications running in physical environments.
Finally, our object-level search and recovery resides at the top layer. Our OEM partnership with Kroll allows us to provide discrete file, object, message, and table-level recoveries for SharePoint, Exchange, and SQL from application-consistent snapshots. If you compare this to native restore methods, you dramatically cut down the amount of time (and money) required by more traditional or native restore methods. By using Rubrik Object Search and Recovery, you forgo having to restore the entire server or database just to recover one email message. Now you have a diverse range of options for recovery. In a DR event, you can recover entire servers or a SharePoint farm all at once using Instant Recovery. In the event of a minor corruption, you can recover a single table for SQL. What if someone accidentally deleted an email? Now you can recover that one message from last night’s backup without having to fail any application over to DR.
How does Rubrik Object Recovery and Search work?
- Just search for the VM and select mount to bring the VM’s application/database directly onto Rubrik. There’s no need to provision any additional production storage.
- Use Rubrik’s Object Recovery software (powered by Kroll) to browse through the source backup data to locate the deleted item. Select the item (email message, inbox, object, table, or log) to restore. You can preview the content before restoring.
- Drag and drop tables, objects and messages into your destination server. You effectively restore items back to your production servers without requiring a database rollback, DR failover or environment reconstruction.
In customer conversations, I like to point out that good data “hygiene” or data governance goes hand in hand with data protection. We tend to treat governance as an afterthought when it should be integral to the data protection discussion.
For example, an employee leaves your company and you’re asked to recover emails from a certain timeframe. How do you extract that from backup today? Do you just turn over complete backup tapes? Do you have the option of exporting the collected messages as a .PST to be shared with your legal team ahead of turning them over to the court? With object-level recovery and search, you’re able to find and collect only what’s necessary and export it if need be.
To learn more, check out our Instant Object-Level Search and Recovery and Taking App-Consistent Snapshots for VMware Environments data sheets. Also, keep an eye on us as we enhance these capabilities in the future and offer more in-depth integrations.