POSIX object storage access.
No staging or special tools required.
5–19x faster data ops.
Accelerated listing, deletion, and migration.
Reduced need for scratch storage.
Lowered costs and simplified infrastructure.

Challenge.

Researchers couldn’t access data in the company’s on-prem object storage system (“ObjStore”) using standard POSIX tools. To run genomics workflows, they had to stage data into a high-performance filesystem and manage files with AWS CLI—adding hours of delay.

ObjStore contained critical research data, including large genomics files and directories of small files, making listing, copying, and deleting especially slow. This added friction for users, increased scratch storage costs, and hampered productivity.

Given the scale and sensitivity of their workloads, the company required extensive testing to validate performance, compatibility, and reliability before adopting a new solution.

Solution.

To modernize workflows and remove staging delays, the company adopted Object Mount—a high-performance file interface that provides POSIX access to unmodified objects in S3-compatible storage.

Researchers could now use standard file commands directly on data in ObjStore, with no staging or AWS CLI required. The company conducted thorough testing, validating POSIX compatibility and benchmarking performance for listing, deletion, migration, staging, and bucket-to-bucket copying. They also ran genomics workloads like BWA-MEM directly from object storage.

In every case, Object Mount delivered faster performance, seamless integration, and full compatibility with existing tools and infrastructure.

Result.

With Object Mount, the pharma company dramatically improved workflow speed and usability.

Listing operations were up to 13x faster, and deleting many small files was 5x faster than with AWS CLI—cutting task times from hours to minutes. For large file deletion, performance reached up to 19x faster. Researchers ran genomics workloads directly from ObjStore, eliminating staging delays and even outperforming local high-performance storage.

These improvements reduced scratch storage needs, accelerated time to insight, and drove broader adoption of object storage—without requiring workflow changes or retraining.