Omni Gateway & Multi-Protocol API Management (Course #omni-gateway)
For architects, platform admins, DevOps.
By the end, you can
By the end of this course, you will be able to position and pilot a unified multi-protocol gateway — REST, GraphQL, gRPC, WebSocket, AI endpoints — so heterogeneous backends answer to one control plane, one policy set and one bill.
Running case study
Cascadia Logistics governs 90 APIs spread across a Classic Mule Gateway, ad-hoc NGINX edges and an ungoverned LLM proxy; the latest audit flags inconsistent mTLS and no central traffic view. The platform team must evaluate a unified multi-protocol gateway layer under NFRs: policy rollout under 1 week, mTLS on 100% of east-west calls, per-team LLM spend caps.
Programme every line is something you do
6 modules · 14 topics · in course-flow order
-
Module 1
Gateway generations compared
Objective. Compare gateway generations on deployment models and protocol reach to ground the migration case.
- Compare Classic Mule Gateway, Flex Gateway and Omni Gateway on deployment modes and protocol reach
- Frame migration drivers as measurable gaps: audit findings, policy lead times, per-team cost caps
-
Module 2
One control plane
Objective. Diagram control-plane/data-plane separation and place each runtime within the target topology.
- Sketch control plane vs data plane duties for a hybrid estate mixing Mule and non-Mule runtimes
- Place microservices and AI endpoints in the topology, marking autonomy and failure-isolation boundaries
-
Module 3
Any API, one catalog
Objective. Onboard heterogeneous endpoints under one catalog with uniform ownership and SLA metadata.
- Onboard REST/OAS, GraphQL, gRPC/Protobuf, WebSocket and LLM endpoints with uniform owner/tier/SLA metadata
- Verify health probes per protocol, flagging any endpoint without checks (target: 100% probed)
-
Module 4
Policy once, enforce everywhere
Objective. Standardize mTLS, OAuth and token exchange as reusable bundles validated per protocol.
- Parameterize mTLS, OAuth 2.0 and token exchange as reusable policy bundles across heterogeneous backends
- Prove enforcement with negative tests — expired certs, wrong audiences — on a pre-production gateway
-
Module 5
AI Traffic Economics, Observed at the Edge
Objective. Arbitrate AI traffic costs at the edge — token caps, chargeback, semantic cache — proven by consolidated analytics.
- Rate limit LLM APIs by tokens rather than requests (example ceiling: 100k tokens/min per client)
- Charge back per-team LLM spend with hard stops — alerts at 80% and 100% of budget, 429s carrying actionable headers
- Trial semantic caching on repeated prompts, weighing freshness risk against tokens saved (target: −30% LLM bill)
- Consolidate gateway analytics into Anypoint Insights and propagate trace context across HTTP and WebSocket frames
- Alert on cost spikes, not only latency — a runaway agent shows up in dollars before it shows in milliseconds
-
Module 6
Lab: three backends, one gateway
Objective. Produce the lab artifact: three backend types registered and governed under one gateway configuration.
- Register a Mule app, a third-party microservice and an LLM endpoint under one gateway with shared policies
References & Official Documentation every claim above has a source
Across the course
- Omni Gateway OverviewMuleSoft Docs
Module 1
- Anypoint Platform Gateways OverviewMuleSoft Docs
Module 2
- Omni Gateway Deployment ModelsMuleSoft Docs
Module 3
- Adding an Omni Gateway API InstanceMuleSoft Docs
Module 4
- RFC 8693 — OAuth 2.0 Token ExchangeSpecification
- Transport Layer Security Policy - InboundMuleSoft Docs
Module 5
- LLM Token Based Rate Limit PolicyMuleSoft Docs
Module 6
- Getting Started with Self-Managed Omni GatewayMuleSoft Docs