Software-Defined Networking (SDN) and Cisco SD-Access
SDN Fundamentals Review
- SDN (Software-Defined Networking): An approach to networking that centralizes the control plane into an application called a controller
- Traditional Control Plane: Uses distributed architecture where every network device has its own control plane
- Devices use protocols like OSPF to communicate and share routing information
- Each device has its own ACLs and security rules
- SDN Control Plane:
- Centralizes control plane functions like calculating routes
- Network devices share information with the controller instead of each other
- Controller calculates routes for the entire network
- Can centralize the entire control plane or only part of it
- APIs (Application Programming Interfaces):
- Southbound Interface: Controller uses this to interact programmatically with network devices
- Northbound Interface: Allows administrators to interact with the controller using scripts and applications
SDN Architecture Layers
-
Application Layer (Top):
- Contains scripts and applications that tell the SDN controller what network behaviors are desired
- NOT the OSI model application layer - completely different concept
-
Control Layer (Middle):
- Contains the SDN controller that receives and processes instructions from the application layer
- Contains the centralized control plane of the network
-
Infrastructure Layer (Bottom):
- Contains the actual network devices responsible for forwarding messages across the network
graph TD
A[Application Layer<br/>Scripts & Applications] -->|Northbound API| B[Control Layer<br/>SDN Controller]
B -->|Southbound API| C[Infrastructure Layer<br/>Network Devices]
Things to note: These three layers are specific to SDN architecture and should not be confused with the OSI model layers.
Cisco SD-Access Overview
- Cisco SD-Access (Software-Defined Access): Cisco’s SDN solution for automating campus LANs (office wired and wireless LANs)
- Cisco ACI (Application Centric Infrastructure): Cisco’s SDN solution for data center networks
- Uses spine-leaf architecture extensively
- Cisco SD-WAN: Cisco’s SDN solution for automating WANs
DNA Center
- Cisco DNA Center (Digital Network Architecture Center): The controller at the center of SD-Access
- Hardware: Software application installed on Cisco UCS server hardware
- Dual Roles:
- SDN controller used in SD-Access
- Network manager in traditional networks (without SD-Access)
- APIs:
- REST API for interaction with applications
- Southbound protocols: NETCONF, RESTCONF, Telnet, SSH, SNMP
Things to note: DNA Center can manage both SD-Access networks AND traditional networks, though SD-Access functions only work in SD-Access deployments.
Underlay, Overlay, and Fabric
| Term | Definition | Purpose |
|---|---|---|
| Underlay | The underlying physical network of devices and connections (wired and wireless) that provide IP connectivity | Supports the VXLAN tunnels of the overlay |
| Overlay | The virtual network built on top of the physical underlay network using tunnels | Creates virtual network topology independent of physical topology |
| Fabric | The combination of overlay and underlay (physical + virtual network as a whole) | Complete SD-Access network infrastructure |
Underlay Network Details
- Purpose: Support the VXLAN tunnels of the overlay
- Routing Protocol: Typically uses IS-IS (Intermediate System to Intermediate System) instead of OSPF
- Deployment Types:
- Brownfield Deployment: Adding SD-Access on top of an existing network
- DNA Center will NOT configure the underlay (risk to production network)
- Requires compatible hardware and software
- Greenfield Deployment: A totally new network built specifically for SD-Access
- DNA Center WILL configure devices for optimal SD-Access underlay
- Brownfield Deployment: Adding SD-Access on top of an existing network
Optimal SD-Access Underlay Configuration (Greenfield)
- All switches are Layer 3
- Uses IS-IS as the routing protocol
- All links between switches are routed ports (Layer 3)
- STP (Spanning Tree Protocol) is NOT needed (no Layer 2 loops possible)
- Routed Access Layer: Edge nodes (access switches) act as the default gateway for end hosts
- FHRP (First Hop Redundancy Protocol) like HSRP is NOT needed
graph TD
subgraph Traditional_LAN["Traditional LAN"]
A[End Hosts] -->|Layer 2| B[Access Switch]
B -->|Layer 2 with STP| C[Distribution Switch 1]
B -->|Layer 2 with STP| D[Distribution Switch 2]
C -.->|HSRP Virtual IP<br/>192.168.1.1| A
D -.->|HSRP| A
end
subgraph SD_Access_Underlay["SD-Access Underlay"]
E[End Hosts] -->|Default Gateway| F[Edge Node<br/>Layer 3]
F -->|Layer 3 IS-IS| G[Border Node<br/>Layer 3]
F -->|Layer 3 IS-IS| H[Control Node<br/>Layer 3]
G -->|Layer 3 IS-IS| H
end
Things to note: In traditional LANs, distribution layer switches use FHRP (like HSRP) to provide redundant default gateways. In SD-Access, the access layer switches themselves are the default gateways, eliminating the need for FHRP.
SD-Access Switch Roles
| Role | Function |
|---|---|
| Edge Node | Switches that connect to end hosts (like traditional access layer switches) |
| Border Node | Switches that connect to devices outside the SD-Access domain (e.g., WAN router) |
| Control Node | Switches that use LISP to perform various control plane functions |
Things to note: There is NO “management node” role in SD-Access. Only edge, border, and control nodes exist.
SD-Access Overlay Details
Control Plane: LISP
- LISP (Locator ID Separation Protocol): Provides the control plane of SD-Access
- EID (Endpoint Identifier): Identifies end hosts connected to edge switches
- RLOC (Routing Locator): Identifies the edge switch which can be used to reach the end host
- Maintains a list of mappings of EIDs to RLOCs (DNS-like system)
Data Plane: VXLAN
- VXLAN (Virtual Extensible LAN): Provides the data plane of SD-Access
- Creates tunnels used to actually forward traffic
- The “extensible” in the name indicates support for many different SD-Access features
Policy Control: Cisco TrustSec
- Cisco TrustSec (CTS): Provides policy control such as QoS and security policy
VXLAN Tunnel Operation Example
sequenceDiagram
participant PC1
participant SW1 as SW1<br/>(Edge Node)
participant SW3 as SW3<br/>(Control Node)
participant SW2 as SW2<br/>(Edge Node)
participant PC2
PC2->>SW3: Register: PC2 reachable via SW2
SW3->>SW3: Create EID-to-RLOC mapping
PC1->>SW1: Traffic destined for PC2
SW1->>SW3: Query: How to reach PC2?
SW3->>SW1: Response: PC2 via SW2
SW1->>SW2: Forward over VXLAN tunnel
SW2->>PC2: Deliver traffic
Things to note: Instead of traditional routing tables, LISP uses a DNS-like mapping system to locate destination hosts.
DNA Center Features
Design Section
- Build network hierarchy
- Manage IP addresses and subnets
- Configure DHCP servers, DNS servers
- Map enterprise sites globally with hierarchical organization (by country, etc.)
Policy Section
- Configure group-based access control
- Define policies using source groups and destination groups
- Visual policy grid with color-coded permissions:
- Green: Permitted
- Red: Denied
- Custom: Custom policies
- Engineers can write explanations for each policy
Provision Section
- Manage device inventory
- Add new devices
- Assign devices to specific sites (default: Global site)
- View device status (managed/unmanaged)
Compliance Monitoring
- Check device compliance status
- Monitor software image versions (IOS versions)
- Track security advisories
- Example: Device should run IOS 17.03.03 but currently has 16.11.1c → Non-compliant
Assurance Section
- Monitor network status
- View device health metrics
- Ensure devices are running without issues
DevNet Sandbox: Available at sandboxdnac.cisco.com for hands-on exploration
Intent-Based Networking (IBN)
- Intent-Based Networking: Engineer communicates their intent for network behavior to DNA Center
- DNA Center automatically handles the details of actual configurations and policies on devices
- Benefits:
- Simplifies network management
- Allows engineers to focus on important tasks
- Reduces manual configuration errors
Traditional ACLs vs DNA Center Policies
| Aspect | Traditional ACLs | DNA Center Policies |
|---|---|---|
| Complexity | Can have thousands of entries | Simplified group-based policies |
| Intent Clarity | Intent easily forgotten over time | Engineers can document intent for each policy |
| Deployment | Configure devices one-by-one | Centralized, automatic deployment |
| Error Risk | High (manual configuration) | Low (automated) |
| Management | Distributed, per-device | Centralized on DNA Center |
Example Policy Configuration:
- Source: Developers group → Destination: Test_Servers group = Permit
- Source: Guest group → Destination: Servers = Deny
- Source: Employees group → Destination: Various = Custom policy
Traditional vs DNA Center Network Management
| Feature | Traditional Network Management | DNA Center-Based Management |
|---|---|---|
| Configuration Method | Devices configured one-by-one via SSH or console | Centrally managed via GUI or REST API |
| Initial Deployment | Manual configuration via console before deployment | Automatic configuration from DNA Center |
| Management Style | Distributed, per-device management | Centralized management on DNA Center |
| Policy Storage | Configurations stored on individual devices | Configurations stored and managed on DNA Center |
| Software Management | Manual updates per device | Centrally managed; DNA Center monitors cloud for updates |
| Deployment Speed | Slow due to manual labor | Fast; automatic configuration |
| Error Rate | Higher due to manual effort | Lower; reduced human error |
| Interaction | CLI (SSH, Telnet, Console) | GUI, REST API, NETCONF, RESTCONF, SNMP |
Things to note: Traditional network management isn’t inherently “bad,” but it has potential downsides in terms of time, scalability, and error rates compared to centralized SDN management.
Key Protocols Summary
| Protocol | Layer | Purpose in SD-Access |
|---|---|---|
| IS-IS | Underlay (Control Plane) | Routing protocol for Layer 3 underlay network |
| LISP | Overlay (Control Plane) | Maps EIDs to RLOCs for endpoint location |
| VXLAN | Overlay (Data Plane) | Creates virtual tunnels for traffic forwarding |
| Cisco TrustSec | Overlay (Policy) | Provides QoS and security policy control |
| NETCONF | Southbound API | Configuration protocol for DNA Center to devices |
| RESTCONF | Southbound API | RESTful API for DNA Center to devices |
| SNMP | Southbound API | Monitoring protocol |
| SSH/Telnet | Southbound API | Traditional device access protocols |
Common Misconceptions
- SDN Layers ≠ OSI Model Layers: The SDN Application, Control, and Infrastructure layers are completely different from OSI model layers
- DNA Center is NOT only for SD-Access: It can also manage traditional networks
- SD-Access doesn’t eliminate the underlay: The physical network is still critical; it just uses different protocols (IS-IS instead of OSPF, Layer 3 everywhere)
- STP is not used in optimal SD-Access: All links are Layer 3, so no Layer 2 loops exist
- FHRP is not needed in SD-Access: Edge nodes are the default gateways, eliminating the need for HSRP/VRRP
Key Terms and Acronyms
- SDN: Software-Defined Networking
- API: Application Programming Interface
- DNA Center: Digital Network Architecture Center
- SD-Access: Software-Defined Access
- ACI: Application Centric Infrastructure
- VXLAN: Virtual Extensible LAN
- LISP: Locator ID Separation Protocol
- EID: Endpoint Identifier
- RLOC: Routing Locator
- CTS: Cisco TrustSec
- IS-IS: Intermediate System to Intermediate System
- FHRP: First Hop Redundancy Protocol
- HSRP: Hot Standby Router Protocol
- STP: Spanning Tree Protocol
- IBN: Intent-Based Networking
- ACL: Access Control List
- QoS: Quality of Service
- UCS: Unified Computing System
Commands and Configuration References
While SD-Access is primarily configured through DNA Center’s GUI and APIs rather than traditional CLI, here are key technical specifications to remember:
# DNA Center Access
URL: sandboxdnac.cisco.com
(DevNet Sandbox for testing)
# Southbound Protocols Supported by DNA Center
- NETCONF
- RESTCONF
- SSH
- Telnet
- SNMP
# Example Compliance Check
Expected IOS Version: 17.03.03
Current IOS Version: 16.11.1c
Status: Non-compliant
# SD-Access Underlay Characteristics
- All switches: Layer 3
- Routing Protocol: IS-IS
- Links: Routed ports (no Layer 2 trunks)
- STP: Not required
- FHRP: Not required
- Default Gateway: Edge node (access switch)
Things to note: For the CCNA exam, you don’t need to know specific CLI commands for SD-Access configuration, but you MUST understand the concepts, architecture, protocols, and differences from traditional networking.
Exam Preparation Checklist
✓ Understand the three SDN architecture layers (Application, Control, Infrastructure)
✓ Differentiate between underlay, overlay, and fabric
✓ Know the three SD-Access switch roles (Edge, Border, Control)
✓ Understand LISP’s role (control plane, EID-to-RLOC mappings)
✓ Understand VXLAN’s role (data plane, virtual tunnels)
✓ Know why IS-IS is used instead of OSPF in SD-Access underlay
✓ Explain why STP and FHRP are not needed in optimal SD-Access
✓ Compare traditional vs DNA Center network management
✓ Define Intent-Based Networking
✓ Know DNA Center’s dual role (SDN controller + traditional network manager)
✓ Understand brownfield vs greenfield deployments
✓ Recognize Cisco TrustSec’s purpose (policy control)