In life sciences and pharmaceutical research, time is everything. Whether it’s analyzing genomic sequences or running complex simulations, delays in data access directly slow down innovation.
That’s why one of the world’s top five pharmaceutical companies—a leader in diagnostics and genomics-based research with tens of thousands of employees and a global R&D presence—put Storj’s Object Mount to the test.
Their on-premises object storage system held petabytes of critical research data, but accessing it was slowing down teams across their bioinformatics and data science functions.
Their goal: eliminate friction, reduce delays, and empower researchers with seamless access to unstructured data in their S3-compatible object storage—without replatforming or rearchitecting their workflows.
This blog explores their challenges, how they evaluated Object Mount, and the remarkable results they observed.
The challenge: a POSIX problem in a petabyte world.
For a global leader in pharmaceutical research, every second counts in the race to decode genomic sequences or run high-impact models. But despite their advanced infrastructure, one problem consistently slowed them down: overly complex workflows for working with data.
Their storage infrastructure was solid—but their access model was holding them back.
Research data lived in "ObjStore," a scalable, S3-compatible object storage system. But because it didn’t support POSIX access, researchers couldn’t interact with data using the tools they already knew. Instead, they had to move files into a high-performance local file system before any work could begin.
That core limitation introduced a cascade of downstream challenges:
Workflow delays from staging and copying
Before any analysis could begin, teams had to spend hours staging data from object storage to local disk and exporting results back again. These staging steps were necessary but wasteful—delaying everything from preprocessing to final outputs.
Friction for non-expert users
Not every researcher was fluent in object storage paradigms or comfortable using CLI tools. For those unfamiliar with AWS commands or scripting, even routine file tasks like listing or copying became barriers to productivity.
Underutilized object storage and stalled adoption
Despite significant investment in ObjStore, adoption across teams remained low. Researchers avoided the system due to its complexity, opting instead to duplicate data, create workarounds, or rely on local storage. As a result, the company struggled to realize the full value of its object storage investment.
Overreliance on expensive scratch storage
With data access tethered to local performance filesystems, teams depended heavily on high-cost scratch storage. Scaling these environments meant increasing infrastructure spend—just to maintain workflows that were already inefficient.
Together, these issues created a bottleneck that slowed research, strained infrastructure, and limited collaboration. What the company needed wasn’t just better performance—it was a better way to work with data.
The solution: Testing a better way to work with data.
To tackle their workflow bottlenecks, the pharma company tested Object Mount—a high-speed file system that lets users treat object storage like a local drive.
No staging. No scripting. Just familiar POSIX commands on data in place.
They ran Object Mount through real-world scenarios to see how it compared to their current AWS CLI–based workflows. The goal: find out if simpler access could save time, reduce friction, and unlock the full value of their storage infrastructure.
Spoiler: it did—all without changing tools or moving data.
Test results.
First things first: making POSIX just work.
Before diving into performance testing, the pharma company had a critical requirement that POSIX commands work seamlessly on their object storage.
Once Object Mount was mounted, their S3-compatible storage environment transformed—buckets appeared as virtual directories, and objects as virtual files. This made it possible to use standard file system operations, just like on local disk.
To validate compatibility, the team ran functionality tests using a wide range of common POSIX commands directly on the bucket contents. These included:
Every command executed successfully.
But what stood out most wasn’t just that the commands worked—it was how much faster they were compared to the AWS CLI approach, especially as file sizes and counts increased.
Stress-testing list and delete operations.
Next, the pharma company ran a series of internal benchmarks. They compared Object Mount’s POSIX-based file access against their existing approach using AWS CLI to interact with their on-prem, S3-compatible storage system.
All tests used default settings and reflected real-world workloads:
- Small files: 1,000,000 files, each 1 MB
- Medium files: 1,000 files, each 1 GB
- Large files: 1 file, 1 TB
They started with two of the most fundamental file operations: listing and deleting.
Using Object Mount, the team ran these commands across small, medium, and large file workloads and compared the results to their existing AWS CLI–based workflows. The results were striking:
Listing files
Deleting files
In both listing and deletion scenarios, Object Mount consistently outperformed AWS CLI—by as much as 19× faster for large file deletes. These improvements translated into significantly shorter turnaround times for file management tasks that occur every day in research workflows.
Data migration: turning a painful process into a fast one.
One of the pharma company’s ongoing operational needs was migrating data from their on-prem POSIX-compliant file system into their on-prem object storage. This process was necessary—but notoriously painful. Moving large volumes of research data ate up hours of time and consumed valuable system resources.
With Object Mount in place, the team tested migration performance using fpsync (a parallelized version of rsync) and compared it to their existing method using aws s3 sync.
By combining parallel file sync with direct access to object storage via POSIX, the team reduced migration times from hours to minutes—turning a previously disruptive task into a smooth background operation.
Data staging: faster transfers, even when you still need them.
One of the key benefits of Object Mount is that it eliminates the need to stage data between object storage and POSIX systems—researchers can work directly on the data where it lives. But the pharma company knew real-world adoption takes time, and some teams might still prefer traditional staging workflows during the transition.
So, they benchmarked staging performance in both directions: from object storage to high-performance POSIX filesystems (HPFS) and back again. Using the cp command via Object Mount, they compared results against the aws cp CLI tool.
Even for teams that hadn’t yet shifted to direct POSIX access, Object Mount drastically reduced staging time in both directions—cutting some transfers from over 3 hours to under 20 minutes.
Object duplication: less waiting, more doing.
In some workflows, duplicating data is a necessary step—especially when running tests or analyses on backup copies. For the pharma company, this meant occasionally duplicating large volumes of research data between buckets in their object storage system.
Using cp -r via Storj Object Mount, they saw significant improvements over traditional AWS CLI tools:
These time savings meant researchers could spin up working copies of large datasets much faster—accelerating iteration cycles and reducing the overhead of managing data at scale.
Running workloads directly on object storage.
To push the limits further, the team tested a real-world genomics workload—executing BWA-MEM, a widely used tool for aligning genomic sequences. This time, instead of staging the data onto local storage, they ran the workload directly on files in object storage via Storj Object Mount.
Not only did this eliminate staging entirely—it ran faster than both their local high-performance file system and their in-house genomics platform.
By allowing compute to operate directly on bucketed data, Object Mount simplified the workflow and accelerated it, proving that high-performance doesn’t have to mean high-friction.
Business impact: Significant cost savings.
By removing workflow complexity, Object Mount unlocked a ripple effect of improvements across the organization.
Faster workflows without staging or copying.
Researchers could start working on data immediately. Migration, duplication, and prep tasks dropped from hours to minutes—even in legacy workflows.
Simplified access for non-expert users.
Users could interact with object storage using the tools they already knew, reducing training needs and boosting productivity.
Increased adoption of object storage.
With friction gone, teams that once avoided ObjStore embraced it as their primary data platform. The company finally started seeing ROI on its storage investment.
Reduced dependence on expensive scratch storage.
By eliminating staging, Object Mount cut the need for costly, high-performance scratch space—lowering infrastructure costs and freeing up resources.
Ready to rethink your data workflows?
Object Mount helps teams move faster, simplify infrastructure, and unlock the full potential of object storage—without rewriting a single line of code. Talk to an expert to see how it fits into your environment.
