Design a Clos-based spine-leaf topology sized for a target port count and oversubscription ratio.
Information & Communications Technology
Data Centre Network Fabric Design with Spine-Leaf Architecture
Design a non-blocking spine-leaf data centre fabric built on a Clos topology, covering underlay routing, VXLAN/EVPN overlays, multi-tenant segmentation and the automation needed to operate it at scale.
🗓️Duration5 training days
📚Content4 modules · 8 sessions
🎓On completionAccredited attendance certificate
About the programme
Course Overview
Traditional three-tier data centre networks built around a core, aggregation and access layer struggle with the volume of server-to-server, or east-west, traffic that modern applications generate, and they scale unevenly as pods and racks are added. This course teaches the spine-leaf fabric that has replaced them: a Clos topology where every leaf switch connects to every spine, giving predictable latency and equal-cost multi-path forwarding regardless of where two servers sit. Participants design the underlay routing that makes the fabric work, most often BGP between leaf and spine, then build a VXLAN overlay with an EVPN control plane to stretch layer two segments and provide multi-tenant isolation without the scaling limits of traditional VLANs. The course covers oversubscription ratio planning, anycast gateway design so a server default gateway is available at every leaf, and multi-chassis link aggregation for dual-homed servers. It closes with fabric automation and telemetry, so provisioning a new rack and diagnosing a congested link are both operational tasks rather than one-off engineering exercises.
Expected Learning Outcomes
Configure BGP or OSPF underlay routing that gives every leaf equal-cost paths to every spine.
Build a VXLAN overlay with an EVPN control plane for layer two extension and host mobility.
Design multi-tenant segmentation using VRFs and EVPN route targets to isolate tenant traffic.
Implement anycast gateway addressing so servers reach a local default gateway at any leaf.
Configure multi-chassis link aggregation for dual-homed servers and redundant leaf pairs.
Automate fabric provisioning and use telemetry to identify congestion and failed links quickly.
Who Should Attend
Data centre network engineers designing or refreshing a switching fabric.
Network architects planning a move from three-tier to spine-leaf design.
Cloud and virtualisation teams whose workloads depend on fabric performance.
Network operations staff who provision racks and troubleshoot fabric issues.
Infrastructure engineers designing multi-tenant hosting or private cloud networks.
Solution architects specifying data centre networking for new facilities.
Course Modules
Select any module to see its sessions and points.
01Clos Topology and Fabric Fundamentals
2 sessions · 8 points
Session 1From Three-Tier to Spine-Leaf Design
- Compare traditional three-tier data centre design against a Clos-based spine-leaf fabric.
- Explain why east-west traffic growth drove the move to a non-blocking fabric topology.
- Calculate oversubscription ratio and its effect on cost and available bandwidth per rack.
- Size spine and leaf switch count for a target number of server-facing ports.
Session 2Fabric Redundancy and Scaling to Multiple Pods
- Design leaf redundancy so a single switch failure does not isolate a rack.
- Plan a super-spine layer to interconnect multiple pods without redesigning each fabric.
- Decide port and cabling density per rack based on server and uplink requirements.
- Document a fabric design that records topology, addressing and growth headroom.
02Underlay Routing and Equal-Cost Forwarding
2 sessions · 8 points
Session 1Building the BGP or OSPF Underlay
- Configure point-to-point underlay links between every leaf and spine switch.
- Choose between BGP and OSPF as an underlay protocol based on scale and operational preference.
- Enable equal-cost multi-path forwarding so traffic spreads evenly across spine switches.
- Verify underlay convergence time after a simulated link or spine failure.
Session 2Traffic Engineering and Load Distribution
- Monitor per-link utilisation to detect uneven load across equal-cost paths.
- Tune hashing behaviour so flows, not just packets, are distributed predictably.
- Plan capacity upgrades before an oversubscribed link affects application performance.
- Diagnose a hot spot in the underlay using switch counters and flow telemetry.
03VXLAN, EVPN and Multi-Tenant Overlays
2 sessions · 8 points
Session 1VXLAN Overlay and EVPN Control Plane
- Explain how VXLAN encapsulation extends layer two segments across a routed underlay.
- Configure an EVPN control plane to distribute MAC and IP reachability between VTEPs.
- Design anycast gateway addressing so the same default gateway exists at every leaf.
- Trace a packet from source server to destination through VTEP encapsulation and lookup.
Session 2Multi-Tenancy and Segmentation
- Map tenants to VRFs and EVPN route targets to keep their traffic and routes isolated.
- Apply micro-segmentation policy within a tenant to limit lateral movement between workloads.
- Design shared services access for tenants that need controlled cross-tenant connectivity.
- Validate tenant isolation by testing that unauthorised cross-tenant traffic is dropped.
04High Availability, Automation and Operations
2 sessions · 8 points
Session 1Multi-Homing and High Availability
- Configure multi-chassis link aggregation for servers dual-homed to two leaf switches.
- Design failure domains so a leaf or spine failure affects the smallest possible scope.
- Plan maintenance procedures that drain traffic from a switch before it is serviced.
- Test fabric behaviour under a simulated spine or leaf failure before go-live.
Session 2Fabric Automation and Telemetry
- Automate leaf onboarding with zero-touch provisioning driven by a fabric controller.
- Collect streaming telemetry from fabric switches to monitor latency, drops and buffer use.
- Build alerting that flags a degraded link or switch before it causes an outage.
- Hand over runbooks that cover rack turn-up, fault isolation and capacity expansion.
What the participant receives
4 course modules
A structured syllabus
8 training sessions
across 5 days
32 detailed points
Applied, detailed content
Accredited attendance certificate
On completing the programme
Complete your registration
We will contact you within one business day to confirm.
Ready to start?
Reserve your seat and start building the skill.