Ten years ago Juniper Networks transformed the campus access network by introducing Virtual Chassis to replace large chassis access switches. A silent revolution reverberated through almost every wiring closet soon after and today it is tough to find the big access chassis relics from the past. The change, however, did not make it to the core and distribution layers as the backplane capacity was limited by the packet forwarding engine (PFE) capacity. A lot has changed since then with new PFEs offering multi-terabit capacity, massive forwarding tables and a myriad of tunneling protocols, and enterprises can now deliver the same functions in compact distribution platforms.
Today, Juniper is changing the campus aggregation by launching the new EX4650 enterprise switch with Junos and allowing the transition from 10GbE to 25GbE and 40GbE to 100GbE using a compact 1RU platform. No longer do our customers have to replace line cards to move from 1GbE to 10GbE or 25GbE. The multi-speed ports offer the bandwidth and investment protection with a simple change of a transceiver. Further, 100GbE ports provide the flexibility of using them as 40GbE or 4x25GbE breakout when the network is undergoing a transition. And, did I mention boosting network speed does not necessitate a change in network devices?
Campus networks provide the onramp to multicloud and provide access to the cloud services needed to enable business success and provide best experience to users. It does not matter if the enterprise is using a “cloud first” strategy or relies heavily on a private cloud managed by an expert DevOps team---the EX4650 provides building blocks for an architecture that suits our customers’ needs. Enterprises using the collapsed core/distribution can utilize EVPN-VXLAN based ESI LAG and benefit from a simple scale out network design with Virtual Chassis in the access. The design is a simple extension of the current network, and enables Layer 2 or Layer 3 segmentation and extension using EVPN-VXLAN across the core connecting campus buildings and data centers.
Large enterprises that choose to maintain a best-in-class private cloud want to extend the IP fabric design to the campus. The EX4650 enables these IT teams to design a campus IP fabric with Layer 2 overlay using industry standard EVPN-VXLAN, thus creating a uniform network from campus to data center to multicloud, all while using the same automation through the enterprise.
Does this mean that chassis switches are irrelevant in distribution? No, not quite, but a transition is definitely underway. The level of scale of top of the line silicon like Trio with a programmable microcode based design, massive packet buffers, extremely high forwarding table sizes and choice of architecture that includes EVPN-VXLAN, Junos Fusion or a traditional multi-tier design, makes them useful in many deployments. If our customers are considering an upgrade with chassis based switches, either move them to to an advanced PFE silicon (a la Trio) based design or make way for the new breed of fixed switches.
If, instead, their network upgrade looks like filling an old chassis with new expensive line cards that are restricted by older generation packet forwarding engine or a highly restrictive backplane, then encourage them to adopt the network of the future with the EX4650. For more information.