Results Summary and Analysis

This section contains the KPIs used as solution validation targets. Validated KPIs are multidimensional and reflect our observations in customer networks or reasonably represent the solution capabilities. These numbers do not indicate the maximum scale and performance of individual tested devices. For unidimensional data on individual SKUs, contact your Juniper Networks representative.

The Juniper JVD team continuously strives to enhance solution capabilities. Consequently, solution KPIs might change without prior notice. Always refer to the latest JVD test report for up-to-date solution KPIs. For the latest comprehensive test report, contact your Juniper Networks representative.

Test scenarios include validation of baseline features such as EVPN, EVPN-MPLS, HQoS, BGP flowspec, and unicast RPF on WAN Edge nodes. The functions of SR, MPLS, iBGP and route reflectors are validated on core devices. The test scenarios also covered scaled scenarios with various negative triggers such as FPC/PIC reloads, process restarts, and deactivate/activate of instances.

The JVD topology generates a reasonable multi-vector scale of different features as mentioned below in the tables. The scale reference in Solution and Validation Key Parameters characterizes primary multidimensional KPIs represented in the validated profile. A total of­­ 132 test cases are executed and passed successfully.

Traffic convergence is one of the most critical considerations in a network design. This JVD validates the traffic convergence in different scenarios, such as link/node failure and member-link failures. Table 1 given below includes the JVD results, within the specified latency budgets. Figure 1 illustrates the failure scenarios covered by the JVD. The table summarizes measured network convergence timers.

Note:

The worst-case scenarios are reported in the table.

Table 1: Network Convergence Timers

METRO FABRIC INTRA-AS (milliseconds)
	EVPN-VPWS, ms		EVPN-ELAN, ms		EVPN Type5, ms
EVENT	-	FXC	VLAN-Based	VLAN-Bundled	-
WAN Edge1-P1 link disable	5	10	34	30	16
WAN Edge3-P1 link disable	45	130	104	95	5
WAN Edge4-P1 link disable	28	26	56	27	15
WAN Edge4-P2 link disable	121	115	130	120	35
WAN Edge1 Failure	535	121	1821	2113	3123
WAN Edge3 Failure	1470	1470	1561	2351	3370
WAN Edge4 Failure	600	600	1226	39931	586

Note:

The test case is initiated when all EVPN services traffic flows stuck with WAN Edge 4 node.

Convergence time measurements show that when core-facing links (WAN Edge to P nodes) fail, quick restoration times of an average of ~60ms were observed. This is due to a local protection mechanism (such as LFA) having been implemented, ensuring quick restoration. However, in the event of a WAN Edge node failure scenario, significantly higher convergence times were observed as the global convergence process came into play. Note that the observed convergence time depends on the service scale as the validation is conducted with a total of 2700 EVPN services configured at every node. This figure sits on the higher side compared to most mid-range enterprise-class WAN networks, where we expect the convergence time to be much lower. Overall, these measurements demonstrate the effectiveness of the local protection mechanism and the impact of service scale on global convergence timelines in ensuring network stability and minimizing the impact of outages.

Note:

Contact your Juniper Networks representative for more details on network convergence performance.