The Challenge: Data Growth Doesn’t Follow a Procurement Schedule
Organizations traditionally plan storage infrastructure months or even years in advance. Capacity must be forecast, budgeted, purchased, delivered, installed, and made operational before it can be used. When data grows faster than expected, that model doesn't just create inconvenience; it puts research on the clock.
The university faced exactly this pressure. Its Qumulo environment had grown to about 2 PB in production, and because that data is replicated to a second site for disaster recovery, every petabyte added on the primary side must be matched on the DR side, doubling both the hardware footprint and the cost of any expansion. With production and DR combined, the university's on-premises footprint had reached about 4 PB, and additional research projects and datasets continued to push capacity requirements higher. Without more headroom, the university was facing the real prospect of running out of usable storage, a scenario that could have forced active research projects to pause or slow down while new infrastructure was sourced and brought online.
Expanding physical infrastructure meant starting another hardware procurement cycle, and the timing could hardly have been worse. Supply constraints and longer lead times meant a new purchase wouldn't arrive quickly enough to keep pace with the growth curve. And because the DR requirement doubles with every hardware buy, pricing volatility hits twice as hard: with hardware costs running 100 to 300% higher than in prior cycles, the university faced a potentially massive, unplanned budget hit just to stay ahead of data it had already committed to protecting.
The challenge was not simply finding another place to store data. The university needed a way to expand capacity immediately, without waiting on a procurement cycle it couldn't afford in either time or budget, and without abandoning the environment it had already built.
The university needed an architecture that could:
Add capacity without waiting for new hardware
Relieve pressure on existing on-premises infrastructure before it became a capacity crisis
Avoid doubling every future expansion cost to keep DR in lockstep
Preserve its existing Qumulo investment
Move colder datasets to a more appropriate storage environment
Allow cloud infrastructure to be adopted incrementally
Create a foundation for additional cloud use cases
The Solution: Building a Bridge to Infinite Capacity
Because the university was already running approximately 2 PB of production data on Qumulo, spanning at least two physical sites and replicated for disaster recovery for a total on-premises footprint of approximately 4 PB, Qumulo could provide a natural path beyond that environment's physical limits. The university extended its Qumulo architecture into AWS with an initial 200 TB deployment of Cloud Native Qumulo (CNQ).
The new environment gives the university another destination for data, allowing colder datasets to move from primary on-premises infrastructure into AWS and freeing valuable capacity within the university’s existing environment.
More importantly, Qumulo lets the university treat cloud capacity as an extension of its existing storage strategy. The university can continue using on-premises infrastructure where it makes sense while adding cloud capacity as requirements change.
Instead of choosing between on-premises and cloud, the university can use both.
The university can move data into AWS using Qumulo replication workflows or other migration processes appropriate for the dataset and workload. For many use cases, though, Qumulo’s Cloud Data Fabric (CDF) offers a stronger option: it gives researchers local access to data in CNQ without a full migration, along with capabilities for multi-site collaborative research.
Infinite Capacity Without Waiting for More Hardware
The model changes a fundamental assumption about enterprise storage: adding capacity doesn't always have to start with another hardware purchase.
Physical expansion can involve forecasting demand, purchasing equipment, shipping, racking, configuring, and deploying additional systems before that capacity becomes usable. Extending Qumulo into the cloud gives the university another option. When it needs more capacity, it can expand beyond its data center's physical limits without waiting for another hardware cycle.
That lesson extends well beyond higher education. Data growth is unpredictable across industries. A new project, customer, application, acquisition, production, or dataset can quickly change capacity requirements. Building a bridge to cloud capacity lets organizations respond to that growth while continuing to use the on-premises infrastructure they already own.
Creating a Foundation for Future Cloud Use Cases
The initial deployment gives the university a foundation for expanding its use of cloud infrastructure over time. With CNQ in AWS connected to the two existing on-premises environments, the university can add new workloads as its capacity and data protection requirements evolve. The university has also licensed Qumulo Cloud Data Fabric (CDF), giving campus researchers the option to leverage CNQ for NFS home directories and research datasets.
Disaster recovery is a prime example. Extending data into the cloud creates another option for protecting critical datasets without requiring equivalent secondary infrastructure on-premises, giving organizations another way to evaluate where cloud infrastructure can complement the data center.
Business Impact
Capacity when it is needed
The university can extend its existing Qumulo environment into AWS as requirements change. Instead of purchasing infrastructure months ahead of anticipated demand, the university can gain capacity in minutes, not months.
Less dependence on hardware procurement
Additional cloud capacity provides another option when physical infrastructure cannot be purchased and deployed quickly enough to meet changing requirements.
More value from existing infrastructure
The university can move colder datasets into AWS, helping preserve valuable on-premises capacity while continuing to use its approximately 4 PB on-premises Qumulo footprint, including approximately 2 PB in production replicated for disaster recovery.
Cloud adoption without wholesale migration
The university does not need to choose between its data center and the cloud. It can extend its existing storage strategy into AWS incrementally and expand when it makes operational and economic sense.
Foundation for future cloud initiatives
With Qumulo established in AWS alongside its on-premises environment, and Cloud Data Fabric licensed, the university has a foundation for additional cloud workflows, including data protection, disaster recovery, and enabling campus researchers to leverage CNQ for NFS home directories and research datasets.
The Takeaway: Capacity in Minutes, not months.
This private research university’s story illustrates a challenge that extends across industries. Data growth does not always arrive on schedule, but traditional infrastructure often requires organizations to predict capacity months in advance.
Qumulo provides another option. Organizations can continue getting value from the infrastructure they already own while building a bridge to cloud capacity that can expand as requirements change.
The result is a more flexible approach to growth. Keep the infrastructure that makes sense on-premises, extend into the cloud when you need more capacity, and expand without making every increase in data dependent on another hardware procurement cycle.
That is Infinite Capacity: capacity in minutes, not months.