Rubrik Technology Deep Dive

Arvind Nithrakashyap, co-founder and CTO of Rubrik, presented a deep dive into the technology behind Rubrik during a Tech Field Day session on June 25, 2015. He began by sharing his extensive background in database recovery and scalability from his time at Oracle, where he worked on significant projects like the SuperDome architecture and Exadata. His experience at Oracle, combined with his work at Rocket Fuel, where he built a real-time ad system handling billions of requests daily, laid the foundation for his approach to Rubrik. The core idea behind Rubrik was to bring simplicity and scalability to enterprise backup, inspired by the ease of data recovery in consumer devices like smartphones.
Rubrik's technology stack is divided into three main areas: core services, data management, and interface layers. The core services include a distributed metadata service and a scale-out file system, both designed to ensure fault tolerance and scalability. The metadata service uses Flash for quick response times and stores catalog data, while the file system handles both capacity optimization and on-demand data access. The data management layer oversees the entire lifecycle of data, from ingestion to cloud archiving, and includes a search function for quick file recovery. The interface layer separates hypervisor integration and cloud connectivity, allowing for easy support of various hypervisors and cloud providers through adaptable interfaces.
The presentation highlighted Rubrik's focus on minimizing the impact on primary systems during data ingestion by using Flash to handle multiple data streams efficiently. The system's distributed task framework ensures data locality awareness, scheduling jobs close to where the data resides to reduce network traffic. Rubrik's approach to backup eliminates the need for traditional full backups by using incremental forever snapshots, with the ability to fall back to full snapshots if necessary. The platform's architecture supports seamless scalability, allowing for easy addition of nodes and automatic rebalancing of data. Rubrik also emphasizes the importance of metadata, maintaining a comprehensive vSphere hierarchy to facilitate quick recovery and cloud archiving, ensuring data integrity and security throughout the process.
Recorded at Virtualization Field Day 5 on June 25, 2015. For more information, please visit Rubrik.com or TechFieldDay.co...

[@ShubhamMishra1989]

Is deduplication variable block ? if yes - what is the lowest data block for deduplication. Also, can Rubrik do source based deduplication... ?