CableLabs Specifications Move From De Facto to De Jure
The Merriam-Webster online dictionary defines de facto and de jure as follows:
de facto | di-ˈfak-(ˌ)tō – actual, exercising power as if legally constituted
de jure | (ˌ)dē-ˈju̇r-ē – by right, based on laws or actions of the state
In law and government, de facto describes practices that exist in reality, even though they are not officially recognized by laws, whereas de jure describes practices that are legally recognized, regardless of whether the practice exists in reality.
De facto is commonly used to refer to what happens in practice, in contrast with de jure, which refers to things that happen according to law.
The CableLabs® DOCSIS® series of specifications have been de facto standards for the cable industry for over 20 years. In parallel, CableLabs has contributed its specifications to de jure telecommunications industry standards bodies, specifically the International Telecommunications Union (ITU-T), the European Telecommunications Standards Institute (ETSI), and the Society of Cable Telecommunications Engineers/International Society Broadband Experts (SCTE/ISBE).
In the past, creating a de jure standard was a lengthy process involving the reformatting of the specifications into the standards body’s document template, proofreading the reformatting for any errors, submitting the reformatted content as a contribution, and taking it through the standards body’s standardization process. These steps could take many months or even years. Because CableLabs’ specifications are living documents under strict document control (and may have a certification program linked to revisions under this document control), the official de jure standards body’s copy can quickly become out of date. To keep the copy up to date, the process of reformatting, proofreading, submitting, and going through the standards process, has to be repeated with every revision. This reality has led to the unfortunate result that the official de jure standards have been consistently out-of-date.
However, with the SCTE/ISBE publication of the DOCSIS 4.0 standard, now all three of the relevant de jure telecommunications industry standards bodies simply normatively reference the CableLabs specifications, either directly as is the case with SCTE/ISBE and ITU-T, or indirectly as is the case with ETSI. This simplifies the revision process to one of updating a few normative references and approving them. The process of updating these de jure standards has therefore become a much more streamlined activity and the official de jure standards can remain in sync with the CableLabs specifications.
One might ask, “Why does this matter?” Quite simply, it is a matter of law versus fact, as the definitions of de facto and de jure make clear. Regional governments (“the law”) globally prefer to recognize standards that result from the due process of a de jure standards body rather than a potentially proprietary solution coming from a single manufacturer or industry consortia. The standards process will vet these solutions and typically come with an Intellectual Property Rights (IPR) policy by which all participants must abide. Now that all of the relevant de jure standards bodies have adopted this practice cable operators around the world can confidently purchase products compliant with the CableLabs specifications, knowing that they are also fully compliant with any of the official standards from the aforementioned standards bodies.
In effect, ITU-T, ETSI, and SCTE/ISBE have all recognized that the CableLabs DOCSIS specifications are not only the de facto global standard, but also the de jure global standard. There is no longer is any gap between the facts and the law.
Network and Service Management – The Missing Piece for NFV
Network Functions Virtualization (NFV) enables telecommunications networks to be implemented in software running on high volume industry standard servers as outlined by network operators in a seminal white paper published in 2012. NFV standards have been under development in the ETSI NFV Industry Specification Group since the early part of 2013. The ETSI NFV work provides the foundation for NFV and is being referenced by standards organizations globally, and new open source software communities have sprung up to accelerate NFV implementation. I’ve written about industry progress on NFV in previous blogs but we still have some way to go before NFV is commonplace in telecommunications networks.
The key pieces of NFV, notably Virtual Network Functions (VNFs) run on industry standard compute platforms – basically datacenters; and must be dynamically configured and connected at scale to deliver tangible value; automation is absolutely vital for success. Cloud players such as Amazon and Facebook have mastered automation within the confines of their proprietary datacenters, and as a result their operations require orders of magnitude fewer people. New products and services appear at the speed of code, and customer self-service is taken for granted. Concepts that exploit automation such as Machine Learning are being applied which is supercharging the ability of cloud operators to optimize their systems and create cool new stuff. We in the telecoms industry need to also become masters of automation or we will be left behind in the inexorable march to a software defined future.
While the ETSI NFV Industry Specification Group has worked very hard on the “nuts and bolts” of NFV with a keen eye on automation (in my book the most important benefit of NFV), the industry hasn’t made much headway on the key pre-requisite: automation of the Operations environment. Collaboration to address this essential capability is vitally important for the industry to remain competitive and deliver what our customers need in the future.
Information Modeling and Network Automation
Two very important industry initiatives are underway that will accelerate progress. The first initiative is to harmonize information modeling approaches across the telecoms industry (standards and open source). Unless Standards Development Organizations (SDOs) in the different network domains align their information modeling approaches, network operators will have to deal with an ever increasing degree of complexity as they seek to create new networks and services based on NFV. The second is a new industry-wide effort to foster collaboration on Networks and Service Management.
Towards achieving these goals, in January 2016, CableLabs hosted the first multi-SDO and Open Source workshop on Information Modeling which was widely regarded as the moment when the industry realized the value of harmonization. Aligning Information Modeling approaches is a critical first step to achieving network automation (see the blog by my colleague Tetsuya Nakamura). Information models are the “templates” needed to orchestrate compute resources into a meaningful configuration. In the cloud environment, these templates are used routinely, and we need to use them as well, but unlike cloud operators who work in a proprietary, mostly homogeneous environment, telecoms network operators work in a heterogeneous environment spanning many different network domains and referencing standards coming from many different SDOs. Applying cloud technologies in such an environment is extremely complex. Fortunately, SDOs and Open Source communities have recognized this challenge and an unprecedented era of cross-industry collaboration is getting underway.
Multi-SDO Collaboration is not simple, or it would be routine. The first barrier is the focus of individual SDOs on a narrow domain. Other barriers are culture and modus-operandi, and leadership teams motivated by agenda and timelines specific to their domain. Not to mention the dreaded IPR which can stymie even the most worthy of collaborations.
Second Multi-SDO Information Modeling Workshop
To build and maintain momentum, Deutsche Telekom hosted the second Multi-SDO Information Modelling workshop in Bonn-Germany last month. I co-chaired the event with Klaus Martiny at Deutsche Telekom and Michael Brenner at GigaSpaces, and my CableLabs colleague Tetsuya Nakamura played a key role in organization. The workshop dovetailed with another milestone event, the first cross-industry workshop on Networks and Service Management organized by Deutsche Telekom which addressed the broader challenges for automating telecoms networks.
Participants from the following organizations presented their views on harmonizing information modeling: 3GPP (SA5), ARIA, Broadband Forum, ETSI NFV, IETF, IISOMI, ITU-T, MEF, NGMN, OASIS/TOSCA, ONF, OSM, OPEN-O, ON.Lab/CORD, and TM-Forum.
The discussions were intense and extremely positive. Clearly the spirit of collaboration and a sense of common purpose are as strong now as they were after the CableLabs hosted first workshop which bodes well for maintaining momentum on alignment. Follow-up collaborative activities are structured around a set of key topics which we identified as high priority to be addressed with named owners from different organizations who will be accountable for progress. A public WiKi has been created for anyone to follow progress. Activities include:
- Looking at Federated Information Models as a way to get to a Common Information Model.
- Aligning nomenclature amongst the different organizations in relation to Information Modeling and Data Modeling.
- Collecting Use Cases and Business Requirements as a way to bind the effort towards a practical goal.
- Creating and maintaining central repositories for the numerous information models and data models in use across the industry together with descriptive meta-data and open source tooling.
Achieving harmonization is vitally important for the industry to enable automation of the NFV operations environment so we are setting an aggressive timescale to build momentum through 2017.
What CableLabs is doing in this space
We have a number of activities around NFV and SDN that we are executing on behalf of MSOs. For example, CableLabs is progressing an intensive study of virtualized provisioning of the cable access network to enable programmability, our NFV/SDN reference platform is based on OPNFV and we are looking ahead to support 5G using an end-to-end virtualized architecture that includes low latency edge compute nodes located at the cable head-end. In addition, we are seeking to accelerate NFV/SDN interoperability through our subsidiary Kyrio which has built an interop lab where vendors can work together with operators to validate interoperability for their SDN and NFV solutions.
The NFV journey is only just beginning and 5G will be the first new wave of technology to be designed from the ground up using NFV and SDN technologies. The cable industry, with our low latency access network, is in a leadership position to advance these technologies for the benefit of MSOs and their customers globally.
Cable Information Architecture
Passive Optical Networking – for the Next Generation
Service providers invest billions of dollars in their access networks. Ideally, the deployed technology meets consumer demand for many years, allowing service providers to avoid costly upgrades before fully recovering their investments. In addition to technology longevity, service providers also like to see technology evolution, a next generation, to borrow an overused technology term, to ensure future consumer demands can be met by staying within the same technology family. Nowhere is the next generation moniker more prevalent than in the development of passive optical networking (PON) standards.
Both the ITU-T and IEEE are creating next generation PON standards. The ITU has approved two documents from their G.989 series defining Next Generation PON 2 (NG-PON2), and a third document is currently undergoing final comments. The NG-PON2 architecture settled on a time and wavelength division multiplexing (TWDM) method which stacks four wavelengths in a coordinated manner onto a single fiber, with each wavelength delivering 10 Gb/s. Total NG-PON2 TWDM-PON capacity is therefore 40 Gb/s. Since the ONU only receives one wavelength, the bit rate received by a single ONU is capped at 10 Gb/s. In such a solution, the dynamic bandwidth allocation (DBA) would likely be independent for each wavelength. Multiple vendors have reported demonstrating the TWDM-PON solution. A four-wavelength TWDM-PON is illustrated in the figure below.
Is a 100 Gb/s Solution on the Horizon?
The IEEE 802.3 Working Group has recently formed a Study Group to develop objectives for the next generation of Ethernet Passive Optical Networking (NG-EPON). Several key technology decisions await the NG-EPON Study Group: (1) number of wavelengths, (2) bit rate per wavelength, (3) transceiver tunability, and (4) channel bonding. Faced with the same consumer demands and industry competition as other access network technologies, the NG-EPON Study Group could relatively easily define a four-wavelength, 10 Gb/s-per-wavelength PON to put it on par with NG-PON2, shown above. However, alternative solutions are being investigated by NG-EPON participants that would take PON technology multiple steps beyond, and presumably allow consumer demand to be met for many years into the future. For example, using advanced modulation techniques to provide 25 Gb/s per wavelength, combined with four multiplexed wavelengths on a single fiber, could yield the first 100 Gb/s PON solution. By incorporating channel bonding, a concept popularized by CableLabs and the cable industry in the DOCSIS® 3.0 specifications, an NG-EPON ONU would be capable of receiving one or more wavelengths, potentially receiving 50 Gb/s or more. In a channel-bonded solution the DBA will closely coordinate upstream transmissions on one or more wavelengths simultaneously. Are the optical transceivers tunable? That is another of the many important technology decisions yet to be made. A channel-bonded, time-wavelength division multiplexed PON is shown in the diagram below.
With a keen eye on vendor implementation schedules and interoperability, the increased capacity of NG-EPON ONUs and OLTs could be staged according to a timeline that aligns with consumer demand and service provider requirements, alleviating the need to develop yet another next generation standard (next-next generation?).
These technology decisions will begin to take shape within the NG-EPON Study Group (and subsequent Task Force) starting at the September 2015 802.3 Interim meeting. Anyone with the desire to contribute is welcome to attend. Perhaps the ITU-T will also investigate these same approaches for NG-PON2, ultimately resulting in another step toward a converged optical access solution (See the OnePON blog regarding a converged optical access initiative).
In his role as Vice President Wired Technologies at CableLabs, Curtis Knittle leads the activities which focus on cable operator integration of optical technologies in access networks. Curtis is also Chair of the 100G-EPON (IEEE 802.3ca) Task Force.