Latest Articles
The Cable Metro Packet Optical Transport System

The Cable Metro Packet Optical Transport System

Earlier today at the SCTE CableTec Expo in Denver, Colorado, I had the opportunity to present Juniper’s vision and architecture for a cable oriented packet optical core and metro transport system.   In front of an audience of about 150 cable industry engineers, architects and operators, I spoke of a network in which packets become the foundational technology for a network transport system that enables huge scale, efficiency, elasticity and programmability.   We are calling this the Cable Metro Packet Optical Transport System.   This architecture fits into Juniper’s larger vision for cable, which is an all-packet, end-to-end multiservice delivery platform for advanced cable services and applications.



Modern cable metro and core networks are today built in very fragmented, isolated layers.  Typically there is an optical transport group managing DWDM and ROADM technology, and an IP or IP/MPLS group managing packet routing and switching.  This division of labor results in a less than optimal system, where investment in one layer is sometimes unmatched in another, or bandwidth is allocated in a manner unaligned with the application demands of the network, or the time to scale and implement network upgrades is significantly longer than the ideal.


At Juniper we propose to merge the optical and packet layers into a single entity, creating a fluid, fabric-like transport layer that is programmable and adaptable to the business demands of cable networks.     By leveraging the latest technology in integrated 100 (and future 400 and 1000) Gb/s transponders, advanced silicon switching, and software defined networking, we can construct a network to carry all services that a cable operator is offering (residential and commercial) on one network, in a converged manner, with common management, analytics and traffic engineering processes.


IP networks, as the foundational technology of Internet connectivity, are nearly infinitely flexible and malleable; indeed the any-to-any, statistical multiplexing connectivity model of IP is one of its most appealing attributes.   However, IP by itself can also be unpredictable.   The hop-by-hop lookup nature of IP means there can be a degree of uncertainty in path selection and protection.   Traffic engineering in an IP environment is often very coarse, based on IGP metrics or manual, policy based routing configurations.   Transport networks, by contrast, are very linear and predictable, a known entity, and offer a stable platform upon which to build contemporary networks.


To architect a flexible transport fabric using packet-optical methods, we need the best of both worlds – the any-to-any, statmux benefits of IP with the predictability and management of optical transport.   MPLS provides the necessary tools to build this system.   By applying the services framework that MPLS and MPLS Traffic Engineering provides to the latest in optics, silicon and SDN, we can create a truly differentiated transport network for modern cable networks.   


MPLS provides a mature, proven, standards based framework for implementing policy driven services in packet networks.  MPLS isn’t just for telephone companies or organizations selling circuits.  Rather it provides an abstraction layer and structure for implementing reliable packet based services across any type of network infrastructure.   MPLS Traffic Engineering, when coupled with an external controller known as a Path Computation Engine, creates the ability to construct a full mesh of paths (known as LSP’s, or Label Switched Paths) of infinitely variable size between elements in a network – using policy defined by the network operator and influenced in real time by actual conditions in the network infrastructure.


Contemporary networks often result in stranded bandwidth – bandwidth that is unused because of the point-to-point nature of optical transport.  But MPLS and the packet optical architecture eliminate this stranded bandwidth, resulting in greater efficiency in the network and better throughput for applications.    Pure IP networks often have no ability to create value added services such as VPN’s or ensuring latency sensitive applications take specific paths, or advanced OAM (management) services.   But MPLS natively provides all of this.   MPLS can also carry all IPv4 and IPv6 based high-speed data or residential Internet services.   Indeed, any service offered over this architecture benefits from the statistical multiplexing benefits of packets and the policy creation framework of MPLS TE.


At Juniper we believe the future of cable networks is in packets.   Our vision is that of an all-packet, end-to-end multiservice delivery system for advanced cable applications and services.   This vision extends into the access layer of the network as well.  Working with our partners at Gainspeed, Juniper is developing a modernization of the cable last mile to where all-IP, all-DOCSIS multi-gigabit access infrastructure for both commercial and residential services is a reality.   When realized, the all-packet access network coupled with an all-packet core and metro network will result in a very progressive, flexible and efficient network system for the cable operator.


For more information on our innovations for the cable industry, including the Cable Metro Packet Optical Transport System, visit our web page or reach out to your Juniper cable account team.  We are co-developing this vision with many major cable operators, and we welcome your feedback.




Juniper Employee
The new cable theft of service

The new cable theft of service

Are we going back to the future, inadvertently making it easy to steal cable tv again?


Juniper Employee
Latest Comments
Archive | 06-17-2020
Re: Netconf and YANG – explained in a layman’s term
Archive | 04-09-2020
Re: Per-prefix LFA
By  Knox
Archive | 01-08-2019
Re: Netconf and YANG – explained in a layman’s term