Download the SecureIQLab report
AI and high-throughput data pipeline workloads depend on reliable movement between storage and compute. When S3 data paths slow, fail, or become unstable, model training can stall, RAG pipelines can degrade, and expensive GPU resources can sit idle.
F5 BIG-IP sits in front of Dell ObjectScale to provide a programmable control point for S3 traffic. It helps optimize transport behavior, monitor cluster health, route around failures, and maintain service continuity under adverse conditions.
SecureIQLab validated testing of F5 VELOS and BIG-IP Local Traffic Manager in front of Dell ObjectScale, showing that BIG-IP can be inserted into the S3 data path without compromising throughput when correctly tuned, while adding resiliency during fault and attack scenarios.
Download the report to review the test methodology, architecture, throughput results, and resiliency findings.
Validated by:

How BIG-IP preserves S3 throughput in front of Dell ObjectScale
How BIG-IP maintains live S3 traffic when infrastructure is under stress
Why AI data pipelines need traffic control between storage and compute
The SecureIQLab-validated results show that BIG-IP can provide a resilient delivery tier for Dell ObjectScale environments. In optimized low-latency conditions, BIG-IP introduced only a small insertion cost. In latency sensitive scenarios, BIG-IP removed transport inefficiencies that limited throughput. Under failure and attack conditions, BIG-IP maintained service continuity for live S3 traffic. For AI infrastructure architects, the implication is clear: the storage-to-compute path should be engineered as a managed delivery layer, not treated as a direct connection between clients and object storage.
Note: Performance results are based on testing conducted by F5 in a controlled engineering lab environment and validated by SecureIQLab. Results may vary depending on workload characteristics, infrastructure configuration, TCP profile settings, and network conditions. Report originally published by SecureIQLab in May 2026.