For almost two decades, I’ve been advising clients on their storage and backup and recovery system architectures with solutions that are limited in scalability. The need for scale only becomes more paramount over time and often sneaks up on you. Data generation and consumption grow at exponential rates. Data volume and conversely, the need for scale, is driven by three leading factors:
- Company maturation – Information scales alongside company growth.
- Economies of scale – Economies of scale bring cost savings. For example, Managed Services Providers (MSPs) are aggregators of multiple SMB environments.
- Data explosion – Big Data, mobile cloud era, IoT. Need I say more?
Now try to imagine AT&T Stadium at max capacity (80,000 people) after a Cowboys game, with just one exit hall. We’ll never get out of here. No thank you. This is what it’s like if you’ve been using current backup and recovery solutions available in the market today. I have spent the last few years setting up multiple media agents, servers, global dedupe pools that don’t dedupe across each other because they are silo’d, or even separate resource pools for dedupe data and snapshot data (it’s the same data!). It was the same story over, over, and over again. The software was not designed to manage scale, and the hardware was never architected properly to take on scale. Until now.
Scale is Now Possible
What struck me most about (and why I joined) Rubrik was this notion of unlimited scale. You could start with a small unit of measure and then horizontally scale out by adding more units – both capacity and performance could scale predictably together without an architectural overhaul of the entire solution. We call our units Briks. One Brik typically houses four nodes.
A Cloud-Scale File System, Built from Scratch
Unlimited scalability is possible because our engineers have built a distributed file system from the ground up to manage and store versioned data. One of the leads behind Google Colossus, the file system that makes Google Search possible, led the creation of our Cloud-Scale File System. Our file system can effectively manage data at scale quickly and efficiently – whether this is the initial ingestion of data, application of scale-out global deduplication (no more siloed resource pools), or management of multiple zero-copy clones of a “golden image”.
Now, when I need to manage more data, I can scale out easily by adding more nodes with just a click. I can just as easily downsize my cluster by removing nodes with another click. Managing an ever-expanding cluster doesn’t entail managing separate resource islands. You manage Rubrik as a single system. If you’re in Miami next week, make sure to stop by Nutanix NEXT at the Fontainebleau. I’ll be co-presenting with Matthew Day of Langs Building Supplies on Converged Data Management.