https://sites.google.com/site/routeflow/updates/succesfulldemosatonsnextofeliachangesummerschoolandsupercomputinginseattle

RouteFlow demo at ONS

Lazy to blog about them? To be honest, it may be part of the reason… Priorities change over time, both in personal and professional affairs. Fortunately, the spirit of openness and intellectual sharing is very much alive, only not through this blog (or my underutilized twitter accounts @chesteve @futnetcpqd, more reasons for shame), a fact I intend to turn over. I have at least three unfinished post drafts that should have seen the light… Better never than ever! In this welcome back post I will report on some highlights during the past 12 months. In upcoming posts I will finish and release the existing drafts and provide regular updates on my current activities (if lucky, some of them even with some technical talent) as a Research Scientist at CPqD and happy human being settled in Campinas, Brazil.

What has happened since 07/07/2011?

Me and Carlos Macapuna, the day of Paragliding in Andradas. Looking forward to continue flying high!

OpenFlow is an initiative lead by Stanford University to open up networking devices (switches, routers, access points, base stations) by defining a standard protocol to define the packet forwarding actions. The main abstraction consists of a flow — though the flow table abstraction is still subject to refinements to best expose actual hardware resources — basically formed by any combination among a dozen pieces of information forming the context (i.e., packet header fields up to the transport layer plus incoming port) of a packet to be handled by a device.
You can find more details reading the current specification or the multiple academic papers on OpenFlow. Or even better, put your hands on OpenFlow by following this tutorial (recently held by Stanford people in the major distributed systems Brazilian conference).

OpenFlow is certainly not the first approach to enable some degree of network programmability — the so sought holly grail of network infrastructure providers. Related work can be dated back to the 90´s and the efforts in programming telecommunication networks, the OPENSIG community, IEEE 1520, MPOA (Multi-protocol over ATM), GSMP (General Switch Management Protocol) RFC3292, the active network research thread, and more recently work being done at IETF forces WG (Forwarding and Control Element Separation).

While the motivation behind all this body of work is more or less the same (enabling new features, lowering costs, new revenue streams, etc.). To my understanding, the fundamental difference of the OpenFlow approach is its pragmatism. OpenFlow does not aim at trying to satisfy everybody´s needs and a pragmatic way starts by trying to re-use existing hardware capabilities (e.g, ACL in switches and routers) and defining a simple set of matching rules and associated actions (e.g., forward, discard, send to controller, re-write header XYZ). This way, OpenFlow could be enabled on existing hardware by means of a firmware update. And more importantly, industry attention has been attracted and we can already buy OpenFlow-enabled equipment and prototypes from big players like Cisco, HP, NEC, Extreme, and Juniper are on their way to the product line.

We have called our work QuagFlow and will be presented as a poster in this year´s SIGCOMM edition in New Delhi. See a preview below.

Now that we are progressing with our vision and implementation of QuagFlow, along more literature research, we regret not having cited in our poster version the work by Lakshman et al. at Bell/Lucent on the SoftRouter Architecture. We are coming to a design close to what the SoftRouter has been pursuing on separating control software from routers. Their approach is based on the ForCES protocol vs our OpenFlow interface. Its unclear to me how far their prototypical work has gone along these years. Would like very much to know, especially as we should expect to face a set of similar challenges along our journey.

Further Reading:

• J. Hamilton. Networking: The last bastion of mainframe computing.
• Jacobus Van der Merwe and Charles Kalmanek,
  Network Programmability is the answer!
  What was the question again?
  Or, is there really a case for Network Programmability?
• T.V. Lakshman. T. Nandagopal. R. Ramjee. K. Sabnani. T. Woo.
  The SoftRouter Architecture
• M. Ribeiro Nascimento, C. Esteve Rothenberg, M. R. Salvador and M. F. Magalhães. “QuagFlow: Partnering Quagga with OpenFlow“. To appear in ACM SIGCOMM 2010 Poster Session, Aug. 2010, New Delhi, India.

ABSTRACT

Computing history has shown that open, multi-layer hardware and software stacks encourage innovation and bringcosts down. Only recently this trend is meeting the net-working world with the availability of entire open source net-working stacks being closer than ever. Towards this goal, weare working on QuagFlow, a transparent interplay between the popular Quagga open source routing suite and the lowlevel vendor-independent OpenFlow interface. QuagFlow isa distributed system implemented as a NOX controller ap-plication and a series of slave daemons running along thevirtual machines hosting the Quagga routing instances.

I had the chance to meet very nice people from around the world, have great technical discussions, and of course a great time during the social events and the tourist activities. Sharing experiences with other PhD students has been very fruitful! Being able to hear in person the visions of tier-1 international researchers and getting first hand feedback of your own work …. priceless!

Poster on enabling an information-centric forwarding plane. Discussion with Van Jacobson on CCN.

- More photos from the Trilogy summer school
- And from SIGCOMM09

Now, the event season continues in Brazil. This week we are organizing our First International Workshop on New Architectures for Future Internet. The event will be streamed live and the material from the talks will be posted online. We plan to write and publish a summary report gathering the essence of the discussions and our conclusions for future Internet research activities. Next week, we will have the opportunity to report on Future Internet research at the XXVII Simpósio Brasileiro de Telecomunicações (SBrT 2009).

Joining the trend of using Web 2.0 technologies to make conferences more attractive and useful (with SIGCOMM09 being a remarkable example), we have set up the following information channels. Please feel free to join them. We will offer twitter to let remote participants participate in the Q&A sections and the discussion panels:

- Live streaming of the event
- Twitter futnetcpqd channel for live conversations and remote questions to the speakers
- Linkedin: Join the Future Internet CPqD Open Research Group
- Flickr page with pictures from the workshop

Next time I will post on thoughts about the upcoming events.

You can find most of the presentations and course material online.

I want to point you to the insightful presentation by Van Jacobson on thr work at Xerox PARC on content-centric networking. The slides are online:

Special Invited Plenary Short Course: (CCN) Content Centric Networking by: Van Jacobson

and hopefully the video will be also available. In the large and content-dense talk (around 2,5 hs), Van provided many details of the CCN architecture, presenting the models for content, nodes, routing, security, and transport, unveiling the practical foundations of CCN. He answered many of the questions from the 2006 video, but at the same time opened lots of new questions….

UPDATE 16/07/2010:

Future Internet Summer School (FISS09) Short course

• Index page (see the Wednesday tab)
• Video: part 1, part 2, part 3, part 4

One of my points to take home is that CCN not only works over IP, without new infrastructure requirements, but also very important you could run IP over CCN, by considering host identifiers (IP) as the content-objects you request…

A content layer for the new Internet waist

,

Last week I had the chance to participate in this year’s edition of the IEEE Infocom conferences in the “cidade maravilhosa” of Rio de Janeiro!

Great location, great technical program, and great panel discussion and keynotes. The first day there was an excellent tutorial on sensor networks by Jim Kurose, free of charge thanks to the conference sponsors!

From the panel on Clean Slate Architectures: Where Are We Today, And What Is The Path Forward?

My opinion on clean slate research continues the same: it does not presume clean slate deployment and aims at innovation through questioning the fundamentals withou the constraints of the currently deployed architecture. Recommended introductory reading by Prof. A. Feldman: Internet Clean-Slate Design: What and Why? and my information-centric perspective to future Internet design:

More to take home from Infocom (besides the papers, the promotional books by Pearson, and the calories from the churrascaria “Porcao”):

Talks:

• Nick McKeown, “Software-defined Networking”
• Jim Kurose, “10 Things I Wish My Advisor Had Told Me, ,”

And from my notes of the panel on What Are The Hot Topics in Networking?

Prof. Keith Ross was IMO the one who pointed out the two most interesting trends (I am not that into wireless currently):

1.- The marriage of social networks and P2P: How to enrich / enhance p2p overlays with existing social networks to exploit, proximity, trust and security?

2.- Data center networking: A lot to do to in terms of architecting the large energy-hungry networking infrastructure of data centers. Splitting TCP connections at the front-end, load balancing, intelligent wiring, cost reduction, avoiding bottlenecks (network and I/O), etc… I hope his ppt will be available soon online. Look at SIGCOMM 2008 papers on data center designs to see what kind of research is going on, and more papers in this line are expected in this year’s SIGCOMM.

There was lots of discussion on what are good topics for PhD research. Confirming some of my late suspects, and in contradiction to my previous post on network coding, network coding was considered a topic losing temperature. The practical / killer deployments are not emerging, and for wired scenarios the applications of pure network coding are still yet to be proven effective and efficient to be worth.

For more on cold topics in network: Jon Crowcroft, “Cold Topics in Networking”

and by googling for the link I found a promising material of a recent workshop held by Jon Crowcroft covering research issues and methodology.

Don´t miss his videos on TED if you want to understand:

• the answer on what makes you fell happy or
• why we do not feel miserable if we don’t get what we want,
• is there a “synthetic” happiness,
• the psychological immune system work
• and more…

 Dan Gilbert researches happiness

In a previous post on good content I suggested another video from TED, again one psychologist unveiling our daily human behaviour:

In this post, I just want to share some of the latest books (from multiple disciplines) I have “consumed” and can really recommend:

 Outliers: The Story of Success [AUDIOBOOK] [UNABRIDGED] by Gladwell Malcolm 

An ”outlier” is a super-achiever, like Bill Gates, the Beatles and many others. The author unveils hidden factors that make people extremely successful, with compelling arguments and contradicting the readers’ intuition many times during the nice storyline.  You will find ou how important can be where and when you were born (e.g., premier league athletes mostly born in the first quarter of a year, IT billionaires like Jobs or Gates born around 1955), or why are Asians so good at math (spoiler: it is not in the genes)? This book recalls me another excellent common believe breaker: 

-  Brain Rules: 12 Principles for Surviving and Thriving at Work, Home, and School by John J., Ph.D. Medina

Network Algorithmics,: An Interdisciplinary Approach to Designing Fast Networked Devices (The Morgan Kaufmann Series in Networking) by George Varghese 

 

Patterns in Network Architecture: A Return to Fundamentals by John Day 

 

Even though the practical usage of network coding has yet to be proven in many real networking scenarios, network coding is being considered by major research industry players as part of  the next wave of networking.

The promise of network coding? Gains in terms of network throughput, resilience, security, simplicity… an alternative path to the current practices of boosting network performance, which is basically limited on new networking hardware versions with increased chip rate and memory sizes. 

In my opinion, network coding applied to future inter-networking architectures is an example of research by questioning paradigms and has the potential to introduce another shift in internetworking with an impact comparable to the information theory work of Shannon 60 years ago.

In this post, I will not introduce gratuitous maths and non-rigorous explanation on network coding (please refer to the wast
literature, specially a book for theoreticians and and another one for practitioners). The point I want to make is some key observations that make me believe that network coding is an area worth of exploring for any future networking research project:

• Big computer industry players (e.g., Microsoft, HP, Intel) are investing in network-coding applied research. There may be something ($$$) beyond pure academic research.
• Pioneering research institutions around the world (e.g., Berkeley, MIT) are increasingly publishing the practical results of ongoing research projects.
• Network coding is meeting legacy network settings, e.g., the TCP protocol (Sundajaran et al, INFOCOM ’09). We may see further “transparent” integration of network coding in real systems.

So far, so good. But, network coding is a tricky area. Even though it’s basic concept and the canonical example over a butterfly type of network is pretty simple, the actual field where network coding can be apply and the implementation options is very broad, spanning over all the layers of the traditional network stack:

• Application: Content distribution and P2P systems. The best example is Avalanche by Microsoft Research – related work ). The version applied in real life is called Secure Content Distribution (MSCD) and was trialed recently by to download a new release of Microsoft Visual Studio 2008.  R2:Live Peer-to-Peer Streaming.
• Transport: Network coding meets TCP: A new interpretation of ACKs (revealing newly degrees of freedom). Although, in the paper, coding seems to be E2E, the term NC is used as the coding could be at the end-host or inside the network.
• Network: Practical Network Coding (Chou, 2003)
• Link: XORs in the air, Network Coding forWireless Mesh Networks: A Case Study,    MORE:Trading Structure for Randomness in Wireless Opportunistic Routing
• Physical: Network Coding-based Protection Strategies Against a Single Link Failure in Optical Networks, Symbol-level Network Coding for Wireless Mesh Networks, Efficient, Fault-Tolerant All-Optical Multicast Networks via Network Coding,  Network Coding in a Multicast Switch

I admit a taxonomy based on layers is blury below the network level, where “practical network coding” by Chou et al introduced the notion of mixing packets within generations.

Skepticism is also there (How Practical is Network Coding?). Or should I say good sense (“Mixing Packets: Pros and Cons of Network Coding”). We may assist to some phase of delusion surrounding network coding, if a typical Gartner’s hype cycle can be applied to the field of research.

Question to the community? In which phase would you say is network coding (if research were a technology product)?

1. “Technology Trigger”
2. “Peak of Inflated Expectations”
3. “Trough of Disillusionment”

I have my own list of questions when thinking about the practicality (implementability) of network coding:

• How to decide which information to mix when operating over multiple flows?
• How to carry effciently the coding operations to the final data consumers?
• How to achieve butterfly type of network paths in the information-oriented network under sonsideration?
• Differences and similitudes of network coding over wireless networks compared to wired deployments?
• Feedback channel and applicability in two-way communications (real and non-real time)?
• Security advantages and implications of network coding?
• Interactions with active caching functionalities at multiple levels (packets, pages, documents)?

I will start thinking small, listing the requirements (e.g., multi-paths, identifier space, meaningfulness of packet generation, packet
headers) to perform a strawman approach for network coding in the context of information-oriented networking (e.g., the PSIRP project). Then, we can evaluate the costs and the practical benefits by extensive ns-3 simulations and may be some NetFPGA test implementations.

Whether network coders will eventually supplant routers in large, shared infrastructures like the Internet is very questionable, may be
in the long term as an additional network service… However, I think that we will se more and more real life (niche?) solutions implementing some flavour of network coding. Let it be an IPTV multicast deployment, or Instant Messaging dissemination protocols, error correction algorithms, switch designs or new variants of P2Pcontent distribution schemes like Microsoft’s Avalanche…

I am curious if Rudolf Ahlswede of the University of Bielefeld, Germany could imagine the impact of his research back in
2000. In this post, I have raised more questions than answers. Hopefully, I can turn this over during this promising 2009.

To end with, an optimistic quote of network coding experts:

“By changing how networks function, network coding may influence society in ways we cannot yet imagine.”

say FFROS, KOETTER and MÉDARD.

-Ch.

P.D: I found another optimistic (press-type) reference related to the information-oriented research area: PcMag includes the Van Jacobsen’s content-centric networking (CCN) as one of the “five ideas that will reinvent modern computing“, although I dislike very much the term “Extreme Peer-to-Peer” used by the PcMAg redactors.

P.D2: I could not resist not to google what the blogsphere has also commented on this topic:

- Back to Research: Network Coding and a Small Riddle for You
- Network Coding for Mobile Phones
- Do you know more?
Selected publications on network coding:

• Network Information Flow. R. Ahlswede, N. Cai, S.-Y. R. Li and R. W. Yeung in IEEE Transactions on Information Theory, Vol. 46, No. 4, pages 1204-1216; July 2000.
• T. Ho and D. S. Lun. “Network Coding: An Introduction. Cambridge University Press, Cambridge, U.K., April 2008.
• Fragouli, C. and Soljanin, E. 2007. Network coding applications. Found. Trends Netw. 2, 2 (Jan. 2007), 135-269. DOI= http://dx.doi.org/10.1561/1300000013
• Information Theory and Network Coding by Raymond W. Yeung, The Chinese University of Hong Kong, Springer, August 2008
• Linear Network Coding. S.-Y. R. Li, R. W. Yeung and N. Cai in IEEE Transactions on Information Theory, Vol. 49, No. 2, pages 371-381; February 2003.
• An Algebraic Approach to Network Coding. R. Koetter and M. M?dard in IEEE/ACM Transactions on Networking, Vol. 11, No. 5, pages 782-795; October 2003.
• Polynomial Time Algorithms for Multicast Network Code Construction. S. Jaggi, P. Sanders, P. A. Chou, M. Effros, S. Egner, K. Jain and L.M.G.M. Tolhuizen in IEEE Transactions on Information Theory, Vol. 51, No. 6, pages 1973-1982; June 2005.
• T. Ho, M. Medard, R. Koetter, D. R. Karger, M. Effros, J. Shi and B. Leong. A Random Linear Network Coding Approach to Multicast. In IEEE Transactions on Information Theory, Vol. 52, No. 10, pages 4413-4430; October 2006.
• Jay Kumar Sundararajan and Devavrat Shah and Muriel Medard and Michael Mitzenmacher and João Barros. Network coding meets TCP. CoRR, (abs/0809.5022) 2008. [BibSonomy: dblp] URL
• P. Chou, Y. Wu, and K. Jain, “Practical network coding,” 2003. [Online]. Available: http://citeseer.ist.psu.edu/chou03practical.html
• Barros, J.; “Mixing Packets: Pros and Cons of Network Coding”, Proc Wireless Personal Multimedia Communications Symp. – WPMC, Lapland, Finland, September, 2008.

-Ch.

Today’s use of the Internet arises well known limitations in terms of mobility, security, address space exhaustion, routing and content delivery efficiency. Continuously patching the Internet with ad-hoc protocol extensions and overlay solutions (overlays, CDN, P2P, DPIs, NAT-aware protocols, MIP) is a complex and costly solution for the long term.

Research to circumvent current Internet limitations can be divided into those advocating a completely new architecture (clean-slate), and those defending an evolutionary approach due to incremental deployability concerns. From a research perspective, clean-slate design does not presume clean-slate deployment and aims at innovation through questioning fundamentals [Slide 3 of PSIRP public presentation].

A key question is to what extent a new paradigm thinking ‘out-of-the-TCP/IP-box’ for the future network is really necessary, e.g., as packet switching was to circuit switching in the 70’s. The reasoning is based on the large scale use of the Internet for dissemination of data [Jac06]. Tons of connected devices are generating and consuming content, without caring about the actual data source as long as integrity and authenticity are assured [DONA].

The Internet has shifted from being a simple host connectivity infrastructure to a platform enabling massive content production and content delivery, transforming the way information is generated and consumed. From its original design, the Internet carries datagrams inserted by sending hosts in a best effort manner, agnostic to the semantics and purpose of the data transport. There is a sense that the network could do more and better given that today’s use of the network is about retrieval of named pieces of data (e.g., URL, service, user identity) rather than specific destination host connections. TCP/IP is inherently unfair and inefficient for data dissemination purposes (e.g., multiple flows of P2P applications, redundant information over the wires, etc.). With this in mind, the enhancements of new internetworking layer should not be limited to QoS or routing efficiency: data persistence, availability and authentication of the data itself are beneficial in-network capabilities to be embraced from design [DONA].

Last decade’s efforts towards a next generation Internet, whether clean slate or evolutionary, have mainly focused on end-host reachability, with novel concepts (e.g., id/loc split) addressing the ’classic’ end-to-end security, mobility and routing issues. The common denominator of these proposals is host-centrism.

Research in a new generation Internet has prompted architectural proposals (e.g., FARA, Plutarch, UIP, IPNL, TRIAD, ROFL, NodeID) that mainly aimed at solving the “old” host connectivity and point-to-point communication problems. At the core of these new architectures are more flexible, expressive, and comprehensive naming and addressing frameworks than the Internet hierarchical IP address space.

However, this trend is changing, and senior researchers that have participated in the Internet development since its beginning, have advised to tackle the future Internet problem from an information interconnection perspective.

Van Jacobsen provides a vision [Google video talk] to understand the motivation for a networking revolution; while the first networking generation was about wiring (telephony) and the second generation was about interconnecting wires (TCP/IP), the next generation should be about interconnecting information at large (Content-centric networking) [JAC06]. This shift in the orientation of network architecture design implies rethinking many fundamentals by handling information as a first-class object.

We can also observe this shift toward information-centric networking in the momentum of service oriented architectures (SOA) and infrastructures (SOI), XML routers, deep packet inspection (DPI), content delivery networks (CDN) and P2P overlay technologies. A common issue is the necessity to manage a huge quantity of data items, which is a quite different task than reaching a particular host. In today’s Internet, forwarding decisions are made not only by IP routers, but also by middleboxes, VLAN switches, MPLS routers, DPIs, load balancers, mesh routing nodes and other cross-layer approaches. Moving down data-centric functions to the lower networking layers could be in tune with the trend in access and backbone technologies represented by the coupling of the dominant Ethernet access protocol and label switched all optical transport networks.

Only time will tell whether revolutionary networking concepts get commercially deployed. History has shown that economics and not purely technological arguments is what ultimately turns prototypes into reality. Recent concerning events (and more to come) may potentially promote and accelerate the adoption of new internetworking paradigms.

Our days economy is Internet-sensitive, service outages due to DDoS attacks or due to limitations of BGP insecure routing (remember Pakistan Telecom Youtube shut-down?) carry important worries and expenses: Internet reports claim potential costs of $31.000 per minute for Amazon’s two hour outage in June 2008.

Furthermore, end-users suffer from threats coming from the network such as evolving phishing methods and new forms of SPAM such as SPIT (over IP telephony) or SPIM (over instant messaging), that may end up frustrating the up today so successful Internet-based communication’s experience.

A recent move towards information-centric can be observed in projects addressing the future internetworking mod such as Trilogy, 4ward, EIFFEL, PSIRP, ICT’s FIRE and other activities in the frameworks of EU FP7 and NSF FIND. Similar in spirit, data-centric architectural proposals up today include the DOA, i3, DONA, Haggle and RTFM, in addition to ’peer-to-peer’, ’content-delivery’, ’sensor’ and ’delay-tolerant’ networks.

More than an endless discussion [Darwin] around clean-slate design and actual network (r)evolution deployment, what we really need for the future Internetworking is 1) ‘clean-slate thinking’ beyond the TCP/IP heritage to foster innovation through questioning paradigms; and 2) feasibility work on an information-oriented infrastructure capable of supporting the actual and future demands over the network of networks.

This post is the motivation and background of my current research work [SPSWITCH], now in cooperation with Ericsson Research and the EU FP7 PSIRP project.

References

[JAC06] V. Jacobson. If a clean slate is the solution what was the problem? Stanford ”Clean Slate” Seminar., Feb 2006. [Google video talk]

[RTFM] M. Särelä, T. Rinta-aho, and T. Tarkoma. RTFM: Publish/subscribe internetworking architecture. ICT Mobile Summit, Stockholm., June 2008.

[PSIRP] http://psirp.org

[Darwin] What would Darwin Think about Clean-Slate Architectures?

[DONA] T. Koponen, M. Chawla, B.-G. Chun, A. Ermolinskiy, K. H. Kim, S. Shenker, and I. Stoica. A data-oriented (and beyond) network architecture. SIGCOMM Comput. Commun. Rev., 37(4):181–192, 2007.

[SPSWITCH] C. Esteve Rothenberg, Fabio Verdi and Mauricio Magalhaes. “Towards a new generation of information-oriented internetworking architectures” ACM CoNext, First Workshop on Re-Architecting the Internet (Re-Arch08). Dec. 2008, Madrid, Spain. [PPT] [PDF] [bibtex]