Test Objectives

Test Goals

The testing for this JVDE was performed for the second phase with the following goals in mind. Please also consult the Campus Fabric WAN Router Integration Test Report for more information.

The goals for testing in the first phase were:

• All tests must be performed with a redundant pair of WAN routers like you would deploy in a production-grade environment.
• Test the two major L2 attach functions for stretched VLAN and transport VLAN exit.
• Test the two L3 attach Functions for eBGP-based and OSPF-based exit.
• Test when the WAN router is attached to a core switch as virtual service block function.
• Test when the WAN router is attached to a pair of dedicated service block functions as a physical service block function.
• Test with the following Juniper Devices as WAN router:
  • Juniper MX Series router.
  • Juniper SRX Series Firewall in HA cluster mode.
  • Juniper Smart Session Router as HA cluster.
• Perform failovers simulating a WAN router or service block function outage.
• Perform scale-out tests with 10 VRFs and 500 VLANs distributed in the fabric.

The goals for testing in the current second phase were:

• All tests must be performed with a redundant pair of WAN routers like you would deploy in a production-grade environment.
• The Juniper MX routers used in the previous phase were replaced with Session Smart Routers for this phase. Testing in phase 2 was only performed on SRX Series Firewalls and SSRs.
• Testing was performed with dual stack IPv4/IPv6 networks as overlay as this is a new feature for campus fabric.
• Testing needs to be performed with an IPv6 Network as the underlay as this is new feature for campus fabric.
• The testing will include bridged overlay for EVPN Multihoming fabrics. This is needed for easy migration from legacy MC-LAG environments.
• Phase 1 test cases have not been repeated. Those test cases are stable and the focus is on the new test cases.
• Perform failovers simulating a WAN router or service block function outage.
• Perform scale-out tests with 10 VRFs and 500 VLANs distributed in the fabric.

The table below shows the combination of tests that were executed in each of the two phases.

Layer 2 or 3 integration	Protocol used	Peering option	Fabric underlay	Fabric overlay	MX Router unmanaged	SRX Cluster unmanaged	SRX Cluster Mist managed	SSR HA Mist managed
Layer 2	static	Stretched VLAN	IPv4	IPv4	Phase1	N/A	N/A	N/A
Layer 2	static	Transport VLAN	IPv4	IPv4	Phase1	Phase1	N/A	N/A
Layer 3	OSPF	P2P /31v4	IPv4	IPv4	Phase1	N/A	N/A	N/A
Layer 3	eBGP	P2P /31v4	IPv4	IPv4	Phase1	Phase1 (+ AS Path prepend)	N/A	N/A
Layer 3	eBGP	LAG/ESI-LAG /29v4 peering	IPv4	IPv4	N/A	N/A	Phase2	Phase2
Layer 3	eBGP	LAG/ESI-LAG /29v4 peering	IPv6	IPv4	N/A	N/A	Phase2	Phase2
Layer 3	eBGP	P2P /127v6 + /31v4 peering	IPv6	IPv4+IPv6	N/A	Phase2 (+ AS Path prepend)	N/A	N/A
Layer 2	Bridged Overlay	VLAN (no VRF created)	IPv4	IPv4	N/A	N/A	Phase2	Phase2

Test Non-Goals

The testing for this JVDE was not performed, for various reasons, on the following items:

• No testing with any third-party WAN router or firewall was performed.
• Testing with an SSR WAN router is managed by the Juniper® Session Smart® Conductor.
• Testing bridged overlay in large fabrics because:
  • It is not a good practice as one loses the ability for east-west traffic between VLANs assigned to the same VRF.
  • One loses client visibility as the IP address to MAC address information is only gathered through a fabric VRF.
  • You lose functionality when using VXLAN Group-Based Policy as all blocking functions can only happen then in the same VLAN.