Every cloud environment is rooted in virtualization and is defined by three pillars: network virtualization, server virtualization, and storage virtualization. The VMware NSX Edge Node plays an essential role in virtualizing networking and security services. The throughput supported by the NSX Edge Node is critical for the entire ecosystem and network services running on it.
In this blog, we outline NSX Bare Metal Edge performance for customers implementing Bare Metal Edge for their virtual networking infrastructure. Using NSX Bare Metal Edge (with no services running) with 4x100Gbp interfaces, RFC2544 performance tests yielded a North-South throughput of up to 388 Gbps (97%-line rate) and up to 3 Tbps for the entire cluster (when using 8 Edge Nodes), providing significant throughput for North-South traffic in the virtual network infrastructure.
Please refer to VMware NSX Bare Metal Edge Performance white paper for more information on test and settings used to achieve these results.
The choice of hardware for the NSX Bare Metal Edge is driven by bandwidth requirements and the throughput you want to achieve. Key considerations include:
Today on Network Break we get a plethora of networking news, including Cisco rolling out new custom Ethernet switch ASICs to compete for AI fabrics. Nokia announces new routers also boasting custom silicon, Intel makes noise about the Thunderbolt 5 connector, Marvell touts ASICs for automotive Ethernet, the AfriNIC registry goes into receivership, and more tech news.
The post Network Break 447: Cisco Chases AI Ethernet Fabrics With New Silicon One ASICs; Nokia Announces Routers With Custom Silicon appeared first on Packet Pushers.
Johannes Resch submitted the following comment to the Is Dynamic MAC Learning Better Than EVPN? blog post:
I’ve also recently noticed some vendors claiming that dataplane MAC learning is so much better because it reduces the number of BGP updates in large scale SP EVPN deployments. Apparently, some of them are working on IETF drafts to bring dataplane MAC learning “back” to EVPN. Not sure if this is really a relevant point - we know that BGP scales nicely, and its relatively easy to deploy virtualized RR with sufficient VPU resources.
While he’s absolutely correct that BGP scales nicely, the questions to ask is “what is the optimal way to deliver a Carrier Ethernet service?”
Johannes Resch submitted the following comment to the Is Dynamic MAC Learning Better Than EVPN? blog post:
I’ve also recently noticed some vendors claiming that dataplane MAC learning is so much better because it reduces the number of BGP updates in large scale SP EVPN deployments. Apparently, some of them are working on IETF drafts to bring dataplane MAC learning “back” to EVPN. Not sure if this is really a relevant point - we know that BGP scales nicely, and its relatively easy to deploy virtualized RR with sufficient VPU resources.
While he’s absolutely correct that BGP scales nicely, the questions to ask is “what is the optimal way to deliver a Carrier Ethernet service?”
I’ve just started a new series on network models over at Packet Pushers. The first two installments are here:
I’ve been writing a series about working within the IETF to publish a new standard over at Packet Pushers. The most recent installments are:
Here’s a preview of what I’m working on for those who are interested:
There will probably Continue reading