 | Issue: | Europe II 2014 |
| Article no.: | 1 | |
| Topic: | A vision in the making – A journey towards programmable network | |
| Author: | Günther Ottendorfer | |
| Title: | CTO | |
| Organisation: | A1 Telekom Austria AG | |
| PDF size: | 224KB |
About author
Günther Ottendorfer is CTO of Telekom Austria Group. Günther Ottendorfer started his international career in the telecommunication sector 20 years ago. From 1987 to 1992 the Austrian-born Günther Ottendorfer studied IT at the Technical University of Vienna. 1996 he became a founding member of mobile operator max.mobil in Austria. He gained considerable management experience as COO and CTO for T-Mobile Austria, as CTO for T-Mobile Germany and as European Technology Director at Group level for T-Mobile International.
Since March 2011 Günther Ottendorfer has been a Board Member of Optus Singtel, Australia’s second largest telecommunications provider with over 11 million customers. In addition, he spearheaded the introduction of LTE which Optus pioneered as one of the first providers globally. As of October 2013, Günther Ottendorfer was elected as new Chief Technology Officer (CTO).
Article abstract
SDN disrupting power is changing the landscape, where telecom hardware becomes IT hardware and the network goes to the Cloud. SDN efficiency savings come from sharing computing power, storage and connectivity not only within the Telecom networks but also with the IT infrastructure, leading to organizational merges too. SDN promises an entire new world, where tasks that required highly experienced network engineers will be handled by programmable commands in future. Programmable networks will facilitate easy service creation, even by customers, not by marketing departments and engineering teams. Services will become programmable instead of implementable. The Operator’s role will merely be to facilitate such innovation, as the Internet big players do today.
Full Article
Once in a while, dramatic changes happen in the Telco industry. It seems that such a time has arrived again. There are two relevant technology innovations that promise to change the face of Telecommunications: Network Function Virtualization (NFV) and Software Defined Networking (SDN). In order to make such a change a successful one, many different players will need to work hand in hand, Operators, Telecommunication Equipment Manufacturers (TEMs) and Open Source Communities will be the most important players. Along that journey both, operators and TEMs, need to get ready for a substantial paradigm shift in Telecommunications.
Network Function Virtualization promises to simplify the deployment of network functions by separating the hardware from the software. NFV is therefore about re-architecting the network elements. So far the general view had been that proprietary hardware and purpose-built software is necessary in order to fulfil the performance and availability requirements of Telecom Services. The fact that TEMs have produced and sold HW and SW is proof for that.
Nowadays Operators are looking towards reusing their IT Infrastructure for the deployment of Virtualized Network Functions; the network will become accommodated in a data centre. Additional efficiencies will be generated by sharing computing power, storage space and data centre connectivity between IT and Telecom applications. On the Operator side it will have significant impact on the operations of the network and consequentially the organization. Network and IT operations will now converge to operate a single infrastructure on which software applications are run that serve IT functions as well as Network functions.
TEMs, on the other hand, will need to become software development companies, as the IT HW space is already filled by some large and successful players. IT HW players have a substantial advantage due to their economy of scale. IT systems are relevant not only for Telecom, but practically for any enterprise of today’s world. Industries like banking and trading require substantial IT support that goes in volume of purchased equipment or spent money well beyond the Telecom industry.
This change towards SW-only development companies on TEMs side is facilitated by the fact that vendors have been building layered systems with separation between infrastructure layer and applications for quite some while. That evolution was started with the move towards ATCA equipment and introduction of LINUX operating systems, as well as middleware developed according to Service Availability Forum specifications to achieve high reliability and availability. TEMs are probably better prepared for the transition than operators.
SDN on the other hand promises to simplify the networks themselves. It does so by separating control plane from data plane. At the same time, the control function is being centralized and – similar to NFV – is built on general purpose computing platforms. The centralization provides a better control of the network and the network becomes more flexible and agile in consequence.
SDN is turning the distributed control paradigm upside down. Although forwarding remains a distributed task, calculating the routes through the network is assigned to a single central entity of a domain, which pushes the resulting forwarding tables out to the distributed networking HW. This is nowadays considered as the more flexible and also more agile scheme, allowing the definition of new services and also a simplification of the network management.
Important for this to work are standardized protocols that ensure an abstract view of the underlying network HW. In order for this to be successful commercially, it must ensure that various controllers can interact through the standardized interface with different networking HW. In this way, SDN opens opportunities for new entrants in the Telecom market. Operators hope to profit from this by seeing lower prices that can compensate – at least partially – for the continuous loss of revenues and the continually growing costs of data networks.
SDN promises an entire new world in the networking infrastructure domain, similarly to NFV in the Network Functions domain. Tasks that were resolvable only by highly experienced network engineers will be handled programmatically in the future. A connection between point A and point B in the physical space will mean no more than a configuration entry in the corresponding logical space, i.e. a program ensuring that the corresponding connections will be switched through. NAT or firewall service can be added by a mouse click at the GUI. Establishing connectivity that used to take between three days and several weeks, will be accomplished within a day, may be even within a few minutes.
The promise of SDN for Operators is the ease with which the network can be configured and controlled. New tenant networks can be connected in considerably shorter times and different services can be chained along a single flow, to provide optimum services to subscribers, introducing the potential of customizable services. Another promise brought by SDN is the inherent security, which avoids cross talking between different tenant networks. Once SDN is deployed, Operators hope to gain from automation via programmatic provisioning of the network, as well as the offering of customer self-services via web portals.
Innovation and development will not stop here. Established companies and new market entrants will compete on providing the best network control and optimization SW applications, the best NFV orchestration applications, or the best customer self-service application. Any development will be made available via APIs, so that others programmable network services will be able to reuse them – a paradigm that has been very successful with the large Internet players like Google, Facebook, Amazon, LinkedIn, etc.
Due to the availability of APIs, innovative companies will connect various approaches and services in novel ways for the benefit of customers and the Telecom companies. This in turn will lead to an acceleration of innovation and consequently the services offered to consumers. Big Data applications will run analytics programs across the data obtained from the network and generate information that will be used by programs designed to optimize the consumption and the costs for the Operator. This evolution is probably best summarized by the term ‘Programmable Networks’.
Although ‘Programmable Networks’ encompasses SDN and NFV, it goes far beyond those two technologies. With SDN revolutionizing the network to cope with dynamically changing traffic patterns and traffic growth, and with NFV introducing general purpose computing platforms in data centres and activating all the necessary control functions in a virtual manner, network programmability will allow service creation and service allocation on a dynamic basis. In short, services become programmable instead of implementable .
Such dynamic service creation can be driven by the operator, by third parties or even by customers themselves. Such an approach will support an unprecedented service flexibility and freedom of business models. With the network being a programmable resource, any party will be able to use the network for its vision embodied in innovative services. It is to be expected that the variety of services, that are driven by actual customers’ needs, instead of dedicated engineering, planning or marketing departments, will multiply dramatically.
The advantages of programming services instead of the network are manifold. Operators are no longer subject to the release roadmap of vendors and can use the programmability of their network to come up with services through their own development, test them directly on their market and see the immediate impact. Third parties can realize their own ideas without sharing them beforehand with the operator and easily test them on the market. Where services of the operator are desired, they may be contracted. Customers, if experienced enough with programming (which should be the case for enterprises), can establish their own services or use cases and similarly either contract the operator’s services or work without them. This kind of freedom will foster a plethora of new services with individual price tags. In the background the network warrants the realization of the desired functionality through automation of practically all of the necessary steps in the network and the data centres.
Such an evolution, essentially, will lead to customer defined networks, in contrast to customer supporting networks. The task of the operator will be to make available all those services that a service programmer may need, be it through the Operator or through third party partners.
Although this may sound like fiction today, this will become reality in five to ten years. The ability to make the above happen is driven by the programmability of the networks and the services that are programmed on top of the networks. Their capabilities will be offered in form of APIs, as today’s big players, Google, Amazon and Facebook are offering on the Internet and are asking developer communities to program additional services on top of them.
It seems that Telecom is moving towards an interesting and rewarding future. However, what are the essential guidelines in order to make this evolution a success for the entire Industry?
(1) A level of reliability and availability will continue to play a role. Customers are used to it and will not give up on it. This needs to be observed by all parties involved.
(2) TEMs will select specific core business and focus on that. It does not make sense to try doing everything by oneself and generate closed shop environments that destroy the economy of scale, which is desperately needed by Telecom.
(3) Use and promote Open Source wherever possible. Believe in and foster the power of the many and what they can move, if aligned on a common goal.
