Scale the Access Switching Layer with Junos - dummies

Scale the Access Switching Layer with Junos

By Walter J. Goralski, Cathy Gadecki, Michael Bushong

You may be looking for new ways to scale the access switching layer of your data center to support more servers. If so, you may find that migrating to Juniper Networks access switching solutions is not only cost-effective, but also provides greater flexibility in your network. You can deploy Juniper switches in a variety of access architectures, including top-of-rack or end-of-row, or an innovative combination of both.

Top or end of row

Perhaps you locate your access switches in your data center at the top of the rack (TOR) because it simplifies wiring, keeping more cables within the rack. Yet, many of the small, fixed-configuration switches used in these TOR designs have limited features and functionality. Additionally, administrators must manage each switch individually, increasing operational demands.

Perhaps the access switches in your data center are at the end of the row (EOR) where you can justify larger chassis switches with greater functionality. However, cables must run from all the boxes on the network to the end of the row. If each row doesn’t require all the ports of the large chassis switch, port utilization is low. Also, even a partially configured chassis can have significant space, power, and cooling requirements.

Best of both design options

Ideally, you want an architecture that delivers the benefits of each design option, without their drawbacks. If you had a platform that distributed the ports of a chassis solution across the row or, alternatively, allowed you to incrementally grow at the end of the row as needed, these could be attractive alternatives to your existing design.

Virtual Chassis technology for scaling data center networks

Virtual Chassis technology, available on many Juniper Networks switches enables both of these alternative designs.


For example, the technology enables up to ten Juniper Networks EX4200 Ethernet Switches to interconnect over a 128 Gbps backplane and act as a single logical device sharing a common configuration file. It’s also possible to connect switches located on different floors or in different buildings, using the front-panel 10 Gigabit Ethernet fiber uplink ports.

Deploying Virtual Chassis technology as a part of the access layer solution offers some distinct advantages:

  • Pay-as-you-grow scalability: Expand port densities only as you need, so you can begin economically with a single switch (1 RU) and avoid the up-front investment of chassis-based solutions.

  • Reduced power, cooling, and space needs: Each switch uses far less power and generates far less heat than chassis-based systems, and because you can incrementally add new platforms as you need, you don’t have to pay to power and cool capacity that you aren’t using, or take up valuable rack space with a largely empty chassis.

  • High availability: Interconnect the switches to automatically leverage the multiple Routing Engines to help preserve availability in the rare event of a master switch failure.

  • Location independence: Distribute switches beyond one rack to wiring closets, other floors, and even different buildings, giving you more configuration flexibility. The only distance limitation becomes that of the physical media.

  • Ease of management: Manage and operate up to ten switches as though they were a single physical chassis, simplifying configuration, maintenance, upgrading, and troubleshooting.