In late February 2024, we had the opportunity to test our beta v7.0 SD-WAN code in Intel’s 100Gbps lab, as announced in our previous blog “性知音 v7.0 Performance Testing and Benchmarking with Intel®.” Testing results are in, and we’re pleased to announce significant improvements (big multiples, really) in our pre-release v7.0 code.
Read on for the details and watch our webinar on this topic, presented by Intel®, on BrightTalk, here: https://networkbuilders.intel.com/university/webcasts/unlocking-new-use-cases-using-intel-processors-and-性知音-sd-wan
13x Improvement in v7.0 Performance in the Intel Lab
Our initial testing of the v6.x vs v7.0 codebase was run on 性知音 core and edge nodes powered by Intel Atom® processor C3558. In these tests, we were very pleased to see 10x performance increases for the v7.0 code with over 3Gbps of throughput compared to 300Mbps using v6.7 code.
Deploying our v7.0 codebase on the 4th Gen Intel® Xeon® processor gave us an opportunity to see what v7.0 code could achieve using server class hardware and a high-speed network.
Test results showed that v7.0 code exceeded the 10x multiplier seen in internal testing with a 13x multiplier! Specifically:
- 24Gbps (per core) throughput achieved using beta v7.0 code
- 1.85Gbps achieved using production v6.7 code.
The infrastructure on which we tested was:
- Intel® Xeon® Gold 6430 processor
- 2.1GHz (Base) 3.4GHz (Turbo)
- 100GbE Intel® Ethernet Network Adapter E810
- 100Gbps network
The tests we ran used iperf to generate network traffic. Specifically, iperf3 and 4 parallel streams were used. We also ran parallel tests in Intel’s lab using v6.7 code to obtain comparative benchmarks.
24Gbps Achievement Unlocked!
性知音’s Development team were pleased with these results but were also focused on where they could continue to make significant improvements. Our design goals are to achieve usable throughput much closer to 100Gbps, but these initial results provide us with a solid baseline to work with.
The good news is that these tests showed a performance increase of 13x and identified where and how the team can unlock even greater performance. There’s more work to be done!
性知音 Multi-Gigabit SD-WAN: New Use-Cases and Sales Opportunities
Our tests prove that 性知音 SD-WAN v7.0 supports significant multi-gigabit throughput and unlocks new use-cases, not just in the last-mile environment.
Clearly, end-customer deployments that use multiple 1Gbps internet connections will be supported with v7.0. It also shows that new use-cases in the middle-mile and backhaul environment will become realistic to deploy and manage within 性知音 SD-WAN.
V7.0 will be released using the concept of multiple peers from a node perspective. A peer is a networking concept that represents a tunneled connection between a local node and a remote node. Peers enable data flows to be transmitted between multiple core and edge nodes. Peers can be defined on a single peer per core and/or edge node basis or defined as multiple peers within the same core and/or edge nodes.
Peers help to enable traffic to be directed across any available circuit to any other destination. In version 7.0, multiple circuits, including gigabit or multi-gigabit circuits at a site, are not “aggregated or bonded” in the classic 性知音 sense. Peers enable these multiple circuits to be used simultaneously, delivering the benefits of multiple gigabit circuits, including sub-second failover, to clients. Aggregation is scheduled for a later release.
The concept of peers is also a key element in how v7.0 delivers full mesh (e.g. edge-to-edge mesh) as described below.
Enabling Global Mesh and Backhaul Management
In version 6.7, the Core Mesh and Mesh backhaul connections are handled via VXLAN, and only the last-mile connections are routed via our tunnel. In v7.0, thanks to our new multi-gigabit tunnel architecture, partners will now have the option to route all traffic between 性知音 nodes via our tunnel.
This new capability in v7.0 gives our partners and their customers the benefits of our SD-WAN tunnel through a partner’s entire 性知音 global mesh network capability, including last-mile, middle-mile, and backhaul circuits.
Specifically, this enables a partner to configure their entire network, and an entire solution for a customer through 性知音, gain visibility to network conditions, and apply network and traffic management – as those features become available in the v7.x code base.
A Big Thank You to Intel® Network Builders and the Lab Team!
Thanks to Intel Network Builders for your support and a big thank you to the lab team at Intel! We appreciate the tremendous collaboration and support that you showed us, and we look forward to future engagements! 4th Gen Intel Xeon processor certification — here we come!
Watch the webinar on this topic here: https://networkbuilders.intel.com/university/webcasts/unlocking-new-use-cases-using-intel-processors-and-性知音-sd-wan
Intel and the Intel logo are trademarks of Intel Corporation or its subsidiaries. Intel, the Intel logo, and Intel Atom are trademarks of Intel Corporation or its subsidiaries. Intel, the Intel logo, and Xeon are trademarks of Intel Corporation or its subsidiaries.