By Saverio Romeo, industry manager, telecommunications, at Frost & Sullivan
The network systems that we know today were designed for a client-server world.
Protocols were designed with a “master” in mind from whom passive servants requested bits; in university courses, the analogy of the water pipe systems was often used to describe this scenario.
However, lecturers also advised students that if excessive amounts of water are demanded, the system cannot respond. Similarly, in network systems, if too many bits were requested then the master was in trouble!
The master could add capacity in terms of servers or at the physical layer, then intelligence in the network, maybe at the router level. It could also choose to share the workload with some clients moving towards peer-to-peer engagements.
The essential nature of network systems did not change. It was adapted to the demand of the moment. This adaptation was expensive and took time to develop. Often, the master, despite good intentions, was not ready for the new demand at the time demand was happening.
If some years ago that demand was incremental, today it is radical. The scale of that demand is enormous and growing exponentially. The term “big data” is not misplaced; even less misplaced is its definition using the 3Vs model: volume, variety and velocity.
We are not in front of a water pipe system that requires more pipes and bigger tanks. We are in front of an ocean made of diverse waters that are continuously flowing. And in order to face that ocean, the future network system needs to be flexible and adaptable in almost real time to handle unprecedented different traffic situations.
All of this has to happen with two main criteria in mind: reducing capex/opex and enabling monetisation. Aware of the illness of the current network architectures, the telecommunications industry has started to look at the domain that can develop that flexible, adaptable, and intelligent network system: software.
We should not misinterpret this statement. Software has always been used in networking, but here we are talking about a move – to be emphatic, a paradigm shift – from a hardware-centric networking system to a software-centric one.
If the pipe is not enough, we should not substitute it with a bigger one or add another. Instead, we should enable the network to optimise its performance and ability to supply demand, as well as create condition-aware networks that can predict traffic, recognise and configure new network devices and operate consequently. Only software can provide networks with this intelligence.
To achieve this, the industry is looking into integrating different approaches coming from laboratories such as Software-Defined Networking (SDN) and Network Functions Virtualisation (NFV).
Industry working groups, such as Open Flow and the ETSI Network Functions Virtualizations Group, are playing a key role in this development.
The results coming out from these groups are so promising that, during late 2012 and the first quarter 2013, the industry has come out with SDN and NFV commercial solutions, such as the ones launched by Juniper Networks, Huawei and NEC.
However, this is just the beginning of the software momentum for the networking industry and many questions have already been raised. Software is critical, but the role of big data in making the network adaptable in real-time to conditions is critical as well.
What will the role of big data analytics providers be in this change? But, most importantly, if software and data are the key words for the future of networking, how will the market place change?
Software-centric companies in the networking industry have the right specialisation for the right moment, but what about traditional infrastructure vendors?
Many of them have already embraced the software mantra. Will we have hybrid networking solutions providers (hardware, software and maybe big data)? Will we rather see strong partnerships between players with different expertise?
It is too early for robust answers to these questions, but software-centric networking solutions providers (or ecosystems) will design and dominate the marketplace.
In terms of technological developments, we also believe that all of this will have an extraordinary impact on mobile networks in terms of fifth generation mobile communications research and improvements on existing networks.
The era of the water pipe analogy remains in the book of history of telecommunications and networking. We are heading towards networks that make sense of the diverse entropic states that comprise relentless and chaotic data flow.
The result is the ability to answer demand in the best manner while reducing running costs. Software is the means for this new networking world.