The challenge
A global agricultural R&D division was hitting a wall that had nothing to do with the science itself and everything to do with the infrastructure underneath it. Researchers generating thousands of plant and microbial genomes, alongside a growing volume of phenotyping and trait data, found themselves unable to move that data at the speed their science demanded.
The workaround had become physically shipping hard drives between sites rather than transferring data over the network. For one project alone, that meant splitting 400TB across dozens of drives and manually copying, decompressing, and decrypting each one. Internet download speeds regularly bottomed out at a crawl, sometimes failing outright, thanks to a heavily secured but poorly performing corporate network border designed for office email traffic, not large-scale scientific data transfer.
Compute told a similar story. The organization’s internal HPC cluster was running at roughly 50% utilization, with demand expected to double within a year, pushing it to an unsustainable 100%. A homegrown data analysis and reporting platform had grown so organically over time that its own users were recommending a full rebuild rather than another round of patches. Multiple sites around the world had each built their own version of similar infrastructure independently, with no shared strategy tying them together, duplicating effort and preventing the kind of cross-site collaboration the science actually needed.
Underlying all of it was a cultural tension familiar to any large regulated enterprise: security policy had been designed for business IT, not research IT, and scientists were quietly working around it rather than through it.
The approach
BioTeam ran an onsite assessment combining structured interviews across research and IT staff with a hands-on review of the existing network, compute, and storage environment. Rather than treating the problems as isolated tickets, the point was to connect what researchers were experiencing (slow downloads, failed transfers, and an overloaded cluster) back to root causes in network architecture, security policy, and staffing.
The organization had already taken a first step on its own: a small pilot of a dedicated research network enclave, separate from the main corporate network, with faster internal connectivity. It was a good instinct, but underpowered and available to almost no one. BioTeam’s job was to determine whether that instinct was worth scaling and, if so, how.
The solution
BioTeam recommended building a fully realized Trusted Science DMZ, a dedicated research network operating at high bandwidth (10Gb with a path to 100Gb), interconnecting all research buildings and separated from the general corporate network by purpose-built, high-throughput border equipment rather than office-grade firewalls and proxies. This network would carry a direct, high-speed connection to a major cloud provider for hybrid compute and data sharing, alongside its own dedicated high-bandwidth internet connection for downloading and working with public reference datasets.
Alongside the network redesign, BioTeam proposed:
- A significantly larger, cloud-bursting HPC cluster, sized to handle both current and 3-year projected compute demand, with high-memory nodes for large-scale assembly work and GPU-enabled nodes for workloads that could take advantage of them
- A tiered storage strategy, expanding high-performance parallel filesystem capacity for active data while introducing lower-cost, highly durable object storage and cloud archive tiers for data no longer being actively analyzed
- A full rebuild of the aging internal data platform, redesigned around data commons principles rather than continuing to patch the existing organically-grown system
- A dedicated virtual research support team, split between infrastructure operations and hands-on analytics support, to close the staffing gap that had left mission-critical systems dependent on single points of failure
- A shift toward organization-wide data management practices, including policy-based archival, metadata-driven data curation, and standardized documentation, to replace ad hoc file-and-folder conventions that had made data hard to find and reuse across teams
The outcome
The assessment gave organizational leadership a clear, prioritized blueprint for closing the gap between the scale of science underway and the infrastructure supporting it, with a design built to extend beyond a single site to other research locations globally. Just as importantly, it reframed the conversation happening internally: research IT stopped being viewed purely as a cost to be contained and started being treated as a core, competitive capability to the business’s ability to bring new products to market.
