Redefining application delivery for the AI era with F5 BIG-IP v21 and MCP
In this blog
Artificial Intelligence is no longer confined to research labs or isolated cloud services. Enterprises are rapidly integrating AI models, copilots and autonomous agents into production environments, forcing infrastructure teams to rethink how applications are delivered, secured and managed. Traditional application delivery strategies were built for predictable client-server traffic patterns. AI changes that model completely.
Today's AI ecosystems are dynamic, context-driven, API-heavy and increasingly agentic. They require infrastructure capable of understanding not only traffic flows, but also the context and intelligence exchanged between systems. This is where the evolution of F5 BIG-IP v21, with its MCP-aware functionality, becomes highly significant.
At World Wide Technology (WWT), we see this shift firsthand across enterprise and service provider environments. Organizations are searching for practical ways to operationalize AI securely while maintaining resilience, governance and visibility. BIG-IP v21 introduces capabilities that directly address these challenges, positioning F5 infrastructure as a strategic control point for AI-driven application delivery.
AI is changing the network edge
AI workloads behave differently from conventional enterprise applications. Large language models, AI agents and distributed inference engines generate persistent API interactions, real-time contextual exchanges and massive east-west traffic flows. Security policies designed for static applications often fail to understand the nuances of AI communications.
Understanding Model Context Protocol (MCP)
Model Context Protocol (MCP), introduced by Anthropic in March 2024 and fully supported starting with F5 BIG-IP v21, is an emerging standard designed to enable structured, secure and context-aware communication between AI systems, applications, APIs and orchestration layers.
Rather than treating AI interactions as stateless API calls, MCP enables AI models to exchange contextual information such as memory, task state, intent, operational metadata and tool invocation context. This becomes especially critical in agentic AI environments where autonomous systems coordinate dynamically to complete multi-step workflows.
F5 recognized early that AI traffic requires more intelligent inspection and control mechanisms than traditional Layer 4 or Layer 7 processing. BIG-IP v21 extends the application delivery controller into an AI-aware control plane capable of interpreting MCP traffic patterns and applying granular policy enforcement.
BIG-IP v21: more than a traditional ADC
With v21, F5 evolved BIG-IP beyond conventional load balancing and traffic management. BIG-IP introduced native MCP awareness through programmable iRules, enabling organizations to inspect, process and govern MCP communications at the application delivery layer.
BIG-IP v21.1 advances MCP delivery with native session persistence
BIG-IP v21.1 represents a major operational leap forward through the introduction of MCP Session Persistence as a native Local Traffic Manager (LTM) capability.
MCP workflows are inherently stateful. AI agents and orchestrated LLM workflows depend on maintaining continuity across multiple transactions. If traffic is routed inconsistently across backend systems, session context can be lost, resulting in workflow interruption, failed task execution, or degraded AI outcomes.
While BIG-IP v21 initially enabled session persistence through custom iRules, BIG-IP v21.1 introduces a native MCP Persistence Profile that automatically tracks MCP session identifiers so that all subsequent session traffic is consistently routed to the same backend pool member.
How MCP session persistence works
When an MCP server issues a session ID, BIG-IP v21.1 recognizes the identifier and binds all subsequent requests to the same destination server. This deterministic session affinity preserves contextual integrity across distributed AI environments.
Key benefits include:
* Reliable and resilient AI workflows
* Scale-out support without context loss
* Simplified operational management
* Improved efficiency through reduced context reconstruction overhead
Agentic AI and the rise of autonomous systems
One of the most forward-looking innovations showcased with BIG-IP v21 is support for agentic AI architectures. Agentic AI refers to autonomous AI systems capable of making decisions, orchestrating workflows and interacting with external tools independently.
F5 demonstrated this capability using BIG-IP v21 integrated with MCP and Red Hat OpenShift environments (F5's agentic AI with BIG-IP v21 and MCP on OpenShift). This architecture highlights how BIG-IP can act as a secure policy enforcement and traffic orchestration layer for AI agents operating across containerized infrastructures.
From a networking perspective, this is a major evolution.
Traditional infrastructure was designed around human-driven workflows. Agentic AI introduces machine-driven decision making on a scale, requiring infrastructure capable of real-time contextual inspection, adaptive security policies and intelligent traffic governance. F5 BIG-IP v21 positions itself directly in this operational space.
At WWT, this aligns strongly with what enterprise customers are demanding today: AI enablement without sacrificing operational control. Organizations want the speed and innovation AI provides, but they also require governance, observability and risk mitigation. BIG-IP v21 helps bridge that gap.
Why this matters to enterprise organizations
Many organizations are still in the early phases of AI adoption, but the infrastructure decisions made today will determine how effectively they scale tomorrow.
The challenge is that AI transformation is happening faster than traditional networking and security architectures can adapt. Enterprises need platforms capable of integrating security, application delivery, programmability and AI-aware intelligence into a unified operational model.
WWT's integration role
WWT's AI Readiness Model for Operational Resilience (ARMOR) further amplifies the value of MCP-enabled infrastructure by providing enterprises with a strategic framework for securely operationalizing AI at scale.
As organizations adopt agentic AI, autonomous workflows and distributed inference architectures, MCP becomes a critical mechanism for enabling secure, context-aware communication between models, applications, APIs and orchestration platforms. Within the ARMOR framework, MCP aligns directly with core operational priorities including governance, cyber resilience, observability and secure AI operations. BIG-IP v21 enhances this capability by acting as an intelligent policy enforcement and traffic governance layer capable of inspecting, securing and managing MCP-based interactions in real time.
Combined with WWT's expertise in hybrid cloud integration, automation and AI-ready infrastructure validation, ARMOR provides enterprises with a practical pathway to deploy AI securely while maintaining the operational control, visibility and resilience required for long-term enterprise adoption.
The future of application delivery is context-aware
The networking industry is entering a major transition phase. Application delivery is no longer just about distributing traffic efficiently. It is becoming about understanding intent, context and autonomous interactions between intelligent systems.
MCP represents a foundational shift in how AI systems communicate. BIG-IP v21 positions F5 at the center of this transformation by extending application delivery into the realm of AI-aware governance and contextual security.
For enterprises building next-generation AI platforms, this evolution matters immensely. Infrastructure must now do more than transport packets. It must understand and secure the intelligence embedded within those interactions.
With BIG-IP v21, F5 is redefining the role of the ADC for the AI era. And with WWT helping organizations design, validate and operationalize these architectures, enterprises gain a practical pathway toward scalable, secure and AI-ready digital infrastructure.