Archive for July, 2008

History has shown that business (plus some timing components) and not pure technology is what turns prototypes into reality. In case there are alternatives competing for exactly the same place in the ecosystem, as in nature, only one will survive – and probably not the best from a technology point of view. So far, so good.
But, what if there is no alternative product and a technical solution is killed because it can change the ecosystem, the walled garden? The motivation for asking myself this is a documentary I recently saw and do recommend:

Who Killed the Electric Car?

Don’t ask me where to get it, try your favorite content delivery system.

It hurts to see how good technology enabled by tons of men/day efforts and brilliant ideas can be frozen due to sound business models. Curiously, the electric car is gaining momentum again with the global warming issues and the energy crisis.

I was thinking about similar documentaries that could be done in the computer and network industry, beginning with the historical IBM, OSI, EU vs USA standards battles up to recent high definition video formats (there are always leassons to learn). Of course there are always fair technology wars. Did NAT kill IPv6? If it can be considered dead… We could also me more kind and think of new a new title  like “Who pushed XYZ forward?”.

Talking about good content, free and legal available in the network of networks, I can only recommend you to look at the TED conference 20 minutes talks. Wy pick of today:

Read Full Post »

Recent projects have brought me to dive into the field of the Semantic Web, ontologies and their application to new networking paradigms in both, telco-driven Next Generation Network (NGN) architectures and Next Generation Internet (NGI) architectural proposals. Note that the term “next generation” has become a buzzword that often leads to confusion, since NGN as NGIs have very different focus when looking at future networks.

Telecom world (NGN): In order to make a reality the promise of fast time to market of fancy multimedia blended (prsence+IPTV+FMC+IM+voice+Push-to-X+…) services IMS alone is not enough. The IMS-based service layer of NGNs needs a Service Broker strategy in the Service Delivery Platform to orchestrate the operations involved in service activation, provisioning and finally service execution. SDPs are therefore moving towards service-oriented architectures (SOA) and ultimately towards so called Service Oriented Infrastructures (SOI) – a middleware infrastructure  that natively supports XML processing and all configurable infrastructure resources (compute, storage, and networking hardware and software to support the running of applications). Service Oriented Network Architecures (SONA) is also a term commonly being used for these new event-driven, service-centric architectures, however I dislike it due to the existence of commercial solution by Cisco using this name (nothing at all against Cisco, just prefer to stay with generic terminology).

Time to market of new services over SOA-based SDPs is still too high when you must consider the integration of legacy and heterogeneous OSS/BSS systems of a big telco. An incumbent fixed and mobile operator present in several countries faces the challenge of having heterogeneous service delivery frameworks and trying to deliver the “same service” in the different localities becomes a nightmare due to system heterogeneity and the need of local customization. Not a surprise why major telcos are going the way of outsourcing network operations and opening the service platforms to third parties. We may see up to 100% outsourcing with new business models in a near future, with traditional telcos acting as service supermarkets, putting the infrastructure, branding, auditing the services, billing (! always) and letting service providers the tedious task of managing the SDP and even pushing their products to the front (I hope the analogy is clear, in a future post I will try to explore more this issue).

The main point is that the way telco R&D sees to overcome these heterogeneity issues is not just SOA (XML, Web Services) but it should also include the notion of semantics and the concepts of ontologies (DAML, OWL, RDM). The most relevant effort is the EU SPICE project, and a good paper to understand this approach is:SPICE: Evolving IMS to Next Generation Service Platforms. XML provides standardized data, but only leveraging this data conectors with semantics you are in place to deal with aggregation, adaptation, composition, integration and personalization of heterogeneous systems involved in the service environment. Equipment vendors are already looking on how to incorporate the required semantics and ontologies to implement the Service Orchestration in their SDP solutions, the SCIM in the IMS architecture is a good starting point, but service provisioning and activation in OSS systems are also major elements that could benefit from these semantic enhancements.

Knowledge plain in the SPICE project.
Interestingly, the research coming from the clean-slate wave of future Internet projects (NGI) has also pointed out the requirement of semantics in the new Internet, leading to the notion of a A Knowledge Plane for the Internet. I observed this tendency only recently in a post from Dirk Trossen on the architectural vision called tussle networking (slides here). The EU PSIRP project proposes a new information-centric Internetworking architecture and foresees a kind of knowledge plane based also on the Semantic Web techniques to govern the new networking patterns.

D.Trossen slides

This knowledge plane aims at having a higher-level view of the network. Being an information-aware system that uses artificial intelligence and cognitive techniques to solve the conflicts (tussles) in the network between the different actors (users, network resources, business, organizations). It bases on end-system involvement and is a distributed plane that needs to correlate information from different points in the network. It is supposed to recursively, dynamically and autonomously compose and decompose with the scale of the network. Yes, I also think it may be sound very abstract, futuristic and ambitious, but I am starting to see that this network behavior might be  implementable with enhanced techniques from the Semantic Web. Furthermore, I believe it is the responsibility of network research to think out of the box and try new paradigms, thereby overcoming our often inability to see beyond what is there (this is a topic worth to debate in a future post).

My point is: Once you leverage SOA-based interfaces and database interactions with an ontology-enabled semantic scheme, you are in place to construct this “knowledge plane”, whether for your SDP or for the global Internet – attending in each case the different requirements in terms of performance and scalability. With the gathered mesh of information, the knowledge plane is then in place to behave and perform the reasoning processes to tackle the particular networking tussles.

A practical example coming from the most innovative areas in Internet research in the last decade: CDNs and P2P systems. In my opinion, CDN are already acting somehow as a knowledge agent gathering information about network performance and user location with the goal of redirecting user requests to a concrete (the most convenient) surrogate server. A smart use of this CDN knowledge is a recent P2P project that benefits from the “users vicinity information” granted by the CDN to build the peer links enhancing thus the quality and performance of the overall P2P network.



P.D. Note the similarities around knowledge from the cited works:

NGI perspective: A Knowledge Plane for the Internet:

A network with a knowledge plane, a new higher-level artifact that addresses issues of “knowing what is going on” in the network.

At an abstract level, this is a system for gathering observations, constraints and assertions, and applying rules to these to generate observations and responses.

At the physical level, this is a system built out of parts that run on hosts and servers within the network. It is a loosely coupled distributed system of global scope.

NGN perspective: SPICE Knowledge Services:

Knowledge services provide access to various knowledge gathered from the web, network operators, user profiles and information about local resources.

Local resources are typically discovered by the user terminal and may include nearby devices, accessible networks, local services and sensor data gathered from wireless sensor networks.

The generic form of a knowledge service is called a knowledge source and provides an interface for querying and subscribing to knowledge and register with a knowledge broker.

A knowledge broker is used to find knowledge sources that are able to answer a specific query. Specialized knowledge services include reasoners and recommenders that derive knowledge for personalized end-user services and offer more specific interfaces.

The knowledge layer is in itself a service oriented architecture whose components are used by value added services.

Read Full Post »