Protocols are distributed optimizer

In an amazing paper, F. Kelly, has shown that TCP/IP protocol is a perfect example of optimizer: its objective is to maximize the sum of source utilities (as functions of rates) with constraints on resources. And actually, each variant of congestion control protocol can be seen as a distributed algorithm maximizing a particular utility function. The exact shape of the utility function can be reverse engineered from the given protocol. Similarly, Border Gateway Protocols (BGPs) can be seen as a solution to the Stable Path Problem, and contention-based Medium Access Control (MAC) protocols as a game-theoretic selfish utility maximization. Other utility functions could be User satisfaction (e.g. User-generated pricing following end-to-end principle), resource allocation efficiency or different network economics fairness.

Then, modeling networks as a sets of optimizers, means considering management-control based on interacting controllers maximizing a combination (e.g. weighted sum) or an aggregation (e.g. in multiplicative form) of multiple utility functions. Or, alternatively, we may say that management-control should look for the network Pareto optimality (a state of allocation of resources in which it is impossible to make any one individual better off without making at least one individual worse off: the term is from Vilfredo Pareto (1848–1923), an Italian economist who used the concept in his studies of economic efficiency). But this would mean taking a different perspective in designing and operating networks: i.e., looking for a vertical and horizontal network decomposition and solving problems with multi-objective optimization. In fact, protocols are distributed optimizer !

In a SDN, control intelligence could be (logically) centralized in software-based controllers. Said controllers could provide visibility and control over the network, they can ensure that access control, routing, traffic engineering, QoS, security, and other policies are enforced consistently across the network infrastructures (whilst keeping in mind the CAP Theorem). The interactions of these (centralized vs distributed) controllers will allow managing and optimizing a SDN according to certain policies. Extend this concepts to all layers functions, up to L7 (i.e., consider NFV) and this seems paving the way to look at future networks in a different way, systemically decomposing them and letting ecosystems of resources being aggregated and managed by combinations of multi-objective optimization solvers.

The Network is the Robot

Ultra-broadband diffusion, Information Technology (IT) advances and the market mass decreasing costs are determining a growing adoption of processing and storage resources.

The number of devices, terminals, things connected to the network is growing at an exponential rate: microprocessors and embedded communications are everywhere around us. Smart Cities are going to become populated by any sort of mobile terminals, devices, machines, smart things, sensors, actuators, drones, robots etc.

This is the coming of “machine intelligence“ which will determine a number of socio-economic implications, moving us and the Industry towards the Digital Society and the Digital Economy, where real-time robotics (the new digital tools of mankind) will be pervasive. But robots will require a real-time “nervous system“! 

In this big technology transformation, we need developing an highly flexible and pervasive 5G network, embedding processing and storage capabilities, and capable of providing high bandwidth, ultra-low latency connections. Space-time dimensions of Society and Economy will be morphed.

And a technical challenge will be exploiting the “real-time Operating System“ of this “nervous system“,  i.e., an Operating System intended to serve real-time applications (whatever they are functions or services or related chains) requested from all kinds of terminals (i.e., smartphones, tablets, machines, smart things, cars, robots, drones) attached to the Network and the Cloud. And these borders Terminals-Networks-Cloud will disappear !

So, my take is that controlling and minimizing the “latency” (in its generalised interpretation, including IT response time) will represent a powerful competitive advantage over which building new forms of cooperation and competition, and eventually business models for the Digital Society and Economy. 

If I had to summarize this in a slogan, I'd say: "The Network is the Robot".

Mindset and Skills

Several Bodies, Forum and Communities are arguing that SDN and Virtualization (e.g., NFV) will be two of the key enabling technologies for future networks and services scenarios, even towards the so-called 5G. A wider and wider adoption of IT resources (processing and storage) and software applications (e.g., implementing virtualized network functions) will accelerate the pace of innovation (as it is doing continuously in the IT domains), will reduce costs and will create the conditions for the development of new ICT business ecosystems.

A question is: what are the jobs and the right skills we need to face this evolution ?

It is likely the this IT-zation towards 5G will bring an increase of “management complexity” (the reverse side of greater flexibility and dynamicity) of said infrastructures (in fact virtualization is bringing to a deeper integration of IT and Networks systems on one side, and the Networks with the Terminals on the other side). In order to tame such “complexity” – or better to extract simplicity out of it - it will be necessary, first of all, introducing automated/autonomic operations (e.g., with cognitive, learning methods and systems). A wider adoption of Open Source Software is also expected. 

In my opinion, in order to face said challenges, there will be a need to acquire systemic mindset and to develop/improve the following skills and know how:

1. ability of integrating know how coming from both the IT and Network domains; in the past these two areas grew almost independently, with “client-server” relationships;
2. Industrial Mathematics in order to design and develop systems and methods (e.g., cognitive algorithms, machine learning, autonomic solutions) capable of automating processes; I would say more than the Software it's “Mathematics that will eat the Networks”;
3. “mastering” software (which it is likely to be mainly Open Source) for developing, customizing (and even making testing for performance and validation) of a plethora of SDN or NFV pieces of applications;
4. understanding how handling/correlating the huge amount of data coming from the Network and serive Platforms (e.g., not only apps data but also alarms, nodes states, etc): this includes also skills on Big Data analytics;
5. Cybersecurity and Privacy: in fact IT-zation will exacerbate the security and privacy issues.

If interested, please follow the education activities of our Initiative.

"After Dark" ... The Digital Society

A few days ago I wrote a post about how SDN could meet Cloud Robotics (and other technological domains), arguing that potential promising opportunities are likely to happen around these intersections.

In these days, you might have heard (or see) about the project, called “After Dark”, developed using space exploration technologies. Initiative is about Tate Britain inviting art fans to a night to visit the museum — though robots.

People from around the world can get an after-hours tour on-line thanks to four roaming robots equipped with cameras and sensors and designed to let them move around the Museum rooms.

This is a perfect use case (about cultural heritage) where a robot  can be seen as a particular “Software Defined” – terminal (or an edge node), capable of providing services, being attached to the network (which is also controlling it). This use case is showing several of the challenges under the spot today for future networks: Big Data, 5G, Cognition-as-a-Service… and Users’ involvements, steering new services with their requests. 

Robots roaming an environment, or even actuating actions, are like advanced terminals collecting data (through cameras, sensors…), sending them to a Cloud; a low-latency network is providing robots’ remote control and Users’ access to said services; eventually "cognition" is ensuring the required “intelligence”, leaving this room to competitive "creativity".

A number of similar use cases can be invented and deployed for the Digital Economy and Society !

SDN market estimations

There are a number of estimations of the potential dimensions of the business market of SDN, depending on the metrics adopted by the Analysts. One, for example, is forecasting a $3.52 billion market by 2018, according to Transparency Market Research. Another one is guessing $2.45 billion market by 2018, according to GigaOM Research. So, comparing at least these two reports, we may say a range between $3.5 and $2.4 billion. I've just mentioned two examples, but there are other reports around, obviously.

No need to say that these estimations depend (also) on a number of variables, concerning the introduction scenarios (e.g., Data Center, Core vs Edge vs Home Networks, etc), the adoption rate, the associated business sustainability of such SDN exploitations, and even the regulation aspects. And maybe other important aspects as well.

Nevertheless, as I’ve mentioned in previous posts, SDN should be seen as “network IT-zation”, rather than simply the evolution of networking. This will change the perspective.

We should consider – more broadly - that the network IT-zation is likely to intercept also other trajectories, such as the evolution of the Cloud-Fog Computing and the Internet of Things, enabling new markets for other Industries. Consider, in fact, the number of devices connected to the Telco network is growing at an exponential rate. In a few years the network will become a sort of distributed computational system seamlessly interconnecting a tremendous number of terminals, devices, smart things… robots, drones.

And this could pave the way to scenarios well described in the book the “Second Machine Age”, for which it’s difficult, if not impossible, predicting the business global market today. What is sure, is that this is going to produce a deep impact from a socio-economic perspective, and not only in the Telco Industries.

Have a look at this nice post.

Some challenging questions… about SDN

CAP (Brewer’s) Theorem states that among the three desirable characteristics of a distributed computational system (Consistency, Availability and Partition tolerance), only 2 (but not all 3) can be obtained simultaneously. Given that routing/forwarding packets in a network is a computational problem, then the a network can be seen – from this perspective - as a distributed computational system...

Now, imagine in future SDN scenarios, the availability of programmable and scalable nodes (e.g., routers with computing and storage capabilities), developed starting from commodity hardware and allowing multiple Parties (from Service Providers to end Users) to program, install and execute their services just like network applications. These nodes ideally should be exploited at the edge, as core nodes require by far higher performances.

One may imagine this programmable edge node having multiple instances of execution environment and exposing an Application Programming Interface (API), which is required for the development of any network services applications. Then, one may guess also that there should be a sort of hypervisor capable of receiving each packet, extracting the most important relevant parameters (e.g., MAC/IP addresses, TCP/UDP ports, etc.) and delivering the packet to the execution environment which has to process it to execute some services. Obviously some other services (e.g. the control plane) could be executed on the Cloud.

In principle, all of this can be done with commodity hardware, so with very low cost. But there is one point of attention which has to be considered, in order to get really e-2-e ultra-low latencies: the performance of the execution, especially of of data plane applications (those applications - if any - which will make sense executing on the data plane). I/O operations might limit the expected performances. In fact, data plane packet processing involves moving data from an I/O device to system memory, classifying the data and then moving the data to a destination I/O device. General purpose hardware has been mainly engineered, not for that, but for instruction-bound processing (mainly based on computing instructions rather than I/O), which is mostly local.

If the goal is an ultra-low latency network-service infrastructure, there still unsolved and challenging questions little investigated: where/how executing network services and applications (on the data plane, logically distributed on blades, centralized in the Cloud) ? What’s the required type of Hardware (general purpose, specialized, hybrid) to optimize the costs (including energy consumption) for certain levels of performance ? How handing the CAP Theorem limitations ?

Over-provisioning connectivity pays off better

The throughput of a router is mainly limited by the routing (control plane) processing, which is impacting the maximum number of packets that the router can process at each time: as a consequence there is an inevitable trade-off between the number of ports (node degree) and speed of each port (bandwidth per connection) of  a router. Router Vendors cannot make a router that has both a large degree and a large bandwidth per connection mainly due to the limitation of the routing processing.

Normally nodes in the core network have large bandwidth per connection, and thus small degree.

Vice versa for the edge network: typically the degree of an edge router is almost five times larger than the one of a core router. Even larger if we move nearer the end-Users.

On the other hand, we have to consider that IT advances are making possible to build a 100 (or even more) Gbps software router. Or, software router architectures capable of parallelizing routing functionality both across multiple servers and across multiple cores within a single server (e.g. RouteBricks). 

So, thanks also to SDN and NfV, it is likely it will be possible to build high-speed software routers using low-cost, commodity hardware. This means that it will be possible overcoming routing processing limitation by using the huge amount of processing power made available in large data centers (in other words logically moving the control plane of s/w routers – separated from the forwarding h/w - in the Cloud).

This would change – in principle – the (economic) equation of the network: over-provisioning connectivity rather than just over-provisioning bandwidth. Over-provision connectivity pays off better than over-provision capacity: it is possible creating very large numbers of flexible topologies to choose, even almost randomly, or programming and controlling the QoS to Users' requests. 

Up today, over-provision connectivity in a network is typically more expensive than over-provision capacity, but tomorrow this equation may change. Furthermore, consider also adding to this, the over-provisioning of processing and storage capabilities. That's the IT-zations. It's network adaptability for future ICT services.

In a data center, we have already over-provisioning of connectivity, but the story is different: network covers a relatively small fraction of the cost, compared to server, electricity and cooling costs. So over-provisioning connectivity makes economic sense (by the way, in data centers, traffic demands are quite volatile and not well understood, so it is strictly necessary to over-provision connectivity). Traffic fluctuation on a network has been (up today) over time rather than space, so it has been mitigated by capacity over-provisioning, but this is going to change in the future...