Juniper Networks unveiled its biggest edge router yet today, the M120 multiservice edge router.

Generally available in October, the 120-Gb/s M120 has three times the capacity of Juniper’s next biggest router, the M40e, making it second only to Juniper’s giant M320 core router in terms of scale. In fact, the M120’s 10-Gb/s uplinks give it something in common with the big core router that the other edge routers—all with 2.5-Gb/s uplinks—don’t share. For that reason, Juniper expects the new gear to be useful in small and medium core networks and “collapsed points of presence” (edge services combined with backbone routing) in addition to edge and large enterprise networks.

In designing the new product’s application-specific integrated circuit (which is based on a group of ASICs in Juniper’s big M320 core router), the company focused on two main qualities: scalability and quality of service (QoS).

The chassis includes up to four full-sized “flexible [programmable integrated chip] concentrators” (FPCs), each containing either one or four PICs, and up to two compact FPCs, which require less functionality because they serve uplinks. That functionality trade-off allowed Juniper to pack more capacity into the box.

“[Uplink ports] don’t require things such as being able to go between [virtual local access networks], and they have fewer QoS requirements,” said Alan Sardella, Juniper’s product marketing manager. “They just need to take aggregated traffic coming from other ports and pass it along.”

The initial release of the M120 supports up to 100,000 logical interfaces (which may be virtual local access networks, virtual private networks or any number of different service tunnels). And its Internet routing table can support up to about 1 million routes.

“We’ve left ourselves a great deal of headroom,” Sardella said. “These are industrial, man-sized routing tables.”

The M120 also includes new traffic shaping and policing abilities and double-rewrite functions that help allow carriers to manage end-to-end quality of service specific to the applications that demand it and regardless of the traffic type.

“As you are taking a service end to end and you want to guarantee QoS all the way across multiple networks, you need to be able to enforce that QoS at whatever layer the packets are passing through at a particular time,” Sardella said. “If you’re in the metro and everything’s Ethernet, your class of service bits in the Ethernet frame are important. At some other point, it might be the IP bits or MPLS bits. We break up the packet and make the changes we need to IP or MPLS so that, all the way to the destination, the right class of service will be available.”

One of the M120’s other key attributes is its high-availability. The M120 achieves this with the use of forwarding engine boards (FEBs) that provide redundancy for the FPCs. Customers can choose to have each FPC backed up by a redundant FEB (which provides failover restoration in milliseconds) or, for less protection and a lower cost, one FEB backing up multiple FPCs.

IDC research manager Eve Griliches called the M120 “terrific” in the Juniper press release announcing its introduction. Not only does it deliver “today’s multiservice requirements,” she said, it can support IPTV and triple-play networks as well as converge legacy networks.

Juniper also plans to add the M120’s scalability, high-availability and QoS capabilities to its other products in the future.