Here is a quick recap of the 1st day of ETSI workshop on “future networks” taking place in Sophia Antipolis today.
As usual the agenda was packed with an excellent snapshot of what’s happening and covered some of my personal favorites. Let’s step through them:
Alojz Hudobivnik from Iskratel talked about the importance of early standardization and drew some examples on how old our networks really are and what kind of baggage this implies for standardization in general. It very much reminded me on my recent post on how far telecommunications has come … Alojz looked at the effects of current innovations of cloud computing IoT/M2M and BigData which were concluded with some thought provoking questions about the importance of “thinking it through” at an early stage and asking whether we really want certain features and if they would be good for humans and mankind. I personally thought this was a fantastic philosophical angle which doesn’t always get the attention it deserves during analytical evaluations. Where else could be a better place to bring this up than during a discussion on standards? Awesome!
The next presenter was Dimitri Papadimitriou from Alcatel-Lucent who put some light onto the programmability of networks and whether we need higher level languages to address these issues. Here too I very much enjoyed his “turing church thesis” (–Every effectively calculable function is a computable function) where Dimitri concluded that we are turning 30 year old problems into 80 year old problems. It was an excellent thought angle because when would be a better time for us to look at things with a fresh mind than at the dawn of M2M?IoT !? I can’t think of one!
Andrea Pinnola then rounded up the morning with a view from the operator and a fantastic anecdote on his own children (preferring IP based messaging to SMS) and Beppe Grillo who defies to ballot and campaign rules (eDemocrazy) with using the internet for all his media and PR spectacles. Both entertaining and a very nice aspect from the operators point of view.
The afternoon commenced with a fantastic presentation by Renganai Chaparadza from Fraunhofer FOKUS on customizing autonomic management & control in SDN. It had some excellent ideas on how to make the “knowledge plane” more modular by providing several smarter control loops. The presentation was packed with excellent material which I probably have to study deeper in the coming days in order to get my mind around some of the potential these ideas will create. I very much liked Renganai’s style of presentation because he picked up on what other speakers have said before him and successfully referenced these topics in a very agile manner even remembering the individual peoples names. I liked that!
The next topic was again Renganai who discussed the role of IPv6 in SDN which plays a fundamental role in autonomic set-up and configuration of services.
The next lecture was then given by Pascal Thubert from Cisco who proposed a very interesting solution to routing using an arc concept/topology. His proposal was beautiful because it stayed simple (even I understood how it worked) while at the same time offering both fast-reroute and load-balancing features. Patents are held by Cisco so only question from me to Pascal is now we’ll have to pay and will we ever be able to see support for this in a Linux kernel under GPL license ? 😉
The afternoon was concluded with management in SDN by Alexander Galis from UCL who shined light onto the current limits as well as provided ideas on how the serviceability could be layered in SDN and dividing by Service-Operator-Control, Virtual-Operator-Control, Virtual-Resource-Operator-Control and Physical-Operator-Control as well as how virtualization would fit into the picture.
The last lecture before continuing with a panel show (Q&A) was held by Stefan Wallin from Tail-F in Sweden who talked about the importance of alarms.
Having developed quite a bit of code myself this was one of my favorite topics because it indicated a real need of one of the most overlooked and under-appreciated topics in Software Engineering. Namely the incorrect use and lack of formal specification of what an alarm consists of and when it should be raised. As software developers are charge of a small part of a massive system they rarely get to see the effect of what it means if our code is too chatty. Not only can it degrade performance massively (increased I/O & system calls taking up precious kernel time) but it also misses the point for the customer who rarely cares about the state of the software as long as it does it’s job. Correct handling of alarms is therefore a proper business case and real feature (affecting the bottom line of your products value); while a lack of guidelines can even lead to an effect where the user (operator) might be forced to ignore real errors due to flood of information or simply ignoring certain fault scenarios due to your software “crying wolf” the whole time! The quality of your alarm handling hence is not derived by the massive amount of alarms a system can raise per minute/second but by how intelligently one can fine-tune the level of information that is visible at various scenarios (e.g. during a swap, installation, trouble-shooting or regular operation) and how intelligent this information is presented!
Unfortunately I also missed the best part of the day (networking cocktail) due to a private appointment. However I hope to catch up here tomorrow. Stay tuned 🙂
Valbonne Consulting provides Research & Consulting for emerging technologies in Internet/Web of Things (WoT/IoT/M2M) and Emerging-Tech. We specialise in decentralisation, security and privacy. We work across a variety of traditional industry verticals (Telecommunications, Automotive, Energy, ...). We support Open Source and technologies built on open standards.