A global electronics maker scales AI data delivery with F5

Demand on the S3 object storage of a global consumer electronics manufacturer quickly outgrew what the storage network alone could absorb. F5 VELOS and BIG-IP helped the company keep its AI and analytics pipelines moving to gain faster insight, secure data transport at scale, and room to grow.

Business Challenge

One of the world's largest makers of consumer electronics serves products and customers in markets around the world. It receives telemetry from millions of its devices in the field and uses that data to train its AI models. Its IT organization runs a self-service private cloud that gives internal teams on-demand storage, compute, networking, and security inside the company's own data centers. S3-compatible object storage is one of its foundational services, used by teams across the business.

Like many enterprises rethinking cloud economics in the AI era, the company decided to move large volumes of data out of the public cloud and back on-premises. The biggest source of that data was streaming device telemetry and diagnostic data. Bringing this data in-house gave the company more control and a clear path to lower costs. It also changed the shape of demand on its infrastructure.

Once the data landed on-premises, internal teams moved fast to put it to work. New groups began querying it directly. New data lakes and warehouses appeared. New AI and analytics use cases surfaced almost weekly. S3 storage demand grew to orders of magnitude beyond every other application in the environment, but the company’s private cloud had not been built for that volume and velocity. Pipelines backed up. Analysis and model-training jobs that should have finished in minutes were taking hours. The demands for the data created bottlenecks in the network and storage layer, which meant analysis and training jobs slowed to a crawl. The company needed a data path built to scale.

Solutions

The company chose to deploy F5 hardware in front of its Dell ObjectScale S3 storage to create a resilient, high-throughput data path that could carry the load. F5 solutions and Dell ObjectScale share the same design priorities: high performance, the ability to scale without interrupting operations, and strong support for hybrid environments. F5 products also brought industry-leading throughput, robust SSL/TLS encryption for secure transport, and new S3-specific capabilities built for AI workloads, including the S3 profiles introduced in F5 BIG-IP software version 21.

The team deployed F5 VELOS with high-performance BX520 blades, running F5 BIG-IP Local Traffic Manager (LTM). This gave the company’s S3 traffic a dedicated hardware tier with the capacity to handle multi-gigabit-per-second demand and absorb the spikes that AI workloads create. As storage needs grow, the architecture can be scaled simply by adding blades, with no disruption to live operations.

BIG-IP LTM does more than move traffic. It intelligently distributes S3 object requests across storage nodes to keep throughput high and the service available. It watches for early signs of node stress and rebalances ingestion traffic before a hot spot can stall the pipeline. It also abstracts the storage endpoint, so AI and analytics applications stay connected as the storage scales or changes beneath them. The result is a delivery layer that adds resilience and predictability to the entire S3 environment.

After evaluating other options, the company realized that F5 solutions with new S3 and AI features fit well with Dell ObjectScale and offered the performance, SSL throughput, and future-proof platform needed to empower the organization for the long term.

Results

With a dedicated F5 data path in front of its storage, the company cleared the bottlenecks that had been slowing work. Jobs that once took hours now move through the pipeline far faster. Teams get answers when they need them, and the business keeps pace with demand to answer questions more quickly and make faster decisions.

Turn more telemetry into better AI

The F5 solution gives the company room to retain and process telemetry from a far larger window of time—roughly 30 to 45 days—for AI training, inferencing, and analytics. F5 solutions keep that data flowing so that richer, higher-volume data collection, enrichment, and analysis help the company train better models and produce actionable insights. The company uses those insights to predict device failures before they happen, run operations more efficiently, and optimize shipping and logistics.

Move encrypted data securely at scale

Protecting data in transit was non-negotiable. The F5 solution delivers high-performance SSL/TLS encryption across the deployment, with more than 3,400 Gbps of SSL throughput available to handle growing volumes of encrypted traffic. Sensitive telemetry stays protected without becoming a drag on performance.

Build a platform that grows with the business

Because the VELOS architecture can be scaled without rearchitecture work or downtime, the company can keep adding capacity as new AI and analytics use cases are developed. It can keep building data lakes with confidence that the data path will keep up.

Rely on a tech alliance designed for AI data delivery

This deployment shows what the F5 and Dell technology alliance partnership can deliver. BIG-IP LTM works with Dell ObjectScale to provide resilient, high-throughput S3 data delivery for AI, and the two companies continue to engineer for this exact workload. Recent work includes the S3 profiles in BIG-IP software v. 21, along with joint reference architectures and go-to-market materials that help other organizations scale their own AI data pipelines on a proven design.

For the consumer electronics maker, the relationship is a foundation to keep building upon, as the platform can be efficiently scaled for years as the workload keeps growing.

The lesson for other enterprises is simple. As data volumes and velocity climb with AI demand, a dedicated application delivery tier can turn storage that strains to keep up into a pipeline that is ready to scale.