By Mervyn Kelly, EMEA marketing director at Ciena
In cinema, the concept of ”real-time” has been fairly well defined; 24 frames per second (fps) is, in most cases, enough to fool a human eye into believing that a motion picture projected on a screen is fluid and uninterrupted.
Telecoms has also developed similar real-time benchmarks for the time it takes the network to recover from a failure.
This is referred to as “carrier class” performance – 50 milliseconds for very high performance applications, and sub 300ms for other data related applications.
Historically, the 50ms number was based on the failure-related time limit for circuit switchover after which a voice call could potentially be dropped, causing customer dissatisfaction and potential revenue loss for telecom service providers.
For numerous network applications, such as financial trading or real-time video transfer where any loss of data can potentially have serious consequences, the 50ms benchmark remains very much a tangible beacon of network performance.
Network failure recovery has traditionally been addressed through redundancy.
Ring topology, the prevailing method of optical network build-outs, offers its users a 1+1 protection mechanism, which means that two points on a network ring are connected to each other through two alternative routes.
In the case of failure of the main path (e.g. a fibre break in the network), traffic can be seamlessly redirected onto the protection path.
For smaller networks, especially in metropolitan areas, ring topology remains an effective technique ensuring that networks do fulfil the real-time benchmark.
However, clearly this 1+1 architecture can only survive one failure, and for longer distance networks, the probability of multiple simultaneous failures increases.
One trend that has clearly impacts the situation is globalisation.
With most enterprises operating globally, the world of business relies, more than ever, on international networks to securely interconnect offices, R&D centers, and manufacturing plants for everyday operations.
Large financial organisations operate around the clock and depend on highly secure and trusted global telecommunication infrastructures for conducting millions of transactions a day in world markets.
Cloud services are gaining wide adoption among both service providers and enterprises where resources such as data centers, often geographically dispersed, are remotely accessed by users to perform certain tasks or offer a specific service.
Consequently, global optical networks play an increasingly critical role to the survival, prosperity, and competitiveness of global corporations and the end-users they serve.
Cue intelligent mesh network and optical control plane – the next generation of faster-than-real-time networks that can survive multiple network failures.
In 2011, a large earthquake struck off the east coast of Japan followed by a major tsunami, resulting in extensive damage throughout Japan, including an estimated 12 outages to five separate undersea cables.
In a scenario such as this one, of multiple concurrent failures on a network, the ring protection mechanisms are soon seen waving the white flag of surrender.
That was the also the case in Japan – most carriers’ networks, built out in rings, failed, causing communications disruption at a crucial moment.
The ones that did not were those employing mesh technologies that immediately – often without customers even noticing (or, in cinema language, at 24fps) – rerouted around failures.
Rapid rerouting of connections around inevitable network failures is one of the major benefits enabled by a mesh-based network managed by a control plane.
Properly designed, mesh networks achieve an availability of 99.9999%, which translates into just 31 seconds of downtime annually – that is a ten-fold improvement when compared to traditional ring-based networks.
By virtue of their interconnected topology, mesh-based networks can reduce the required network capacity by over 30 percent in comparison to ring-based networks, as protection bandwidth does not have to be locked up in the 1+1 scenario.
Instead, network intelligence rapidly reroutes connections around multiple failures using its database of available network resources as its reference.
The brain of this intelligent network is its control plane, comprised of hardware and software working together to create a self-aware network that autonomously maintains a database of all its available resources.
Unlike traditional networks which use external intelligence, the control plane essentially creates a network nervous system that controls the underlying network assets and then informs the user of changes performed, resulting in far faster decision-making and connection control.
Already today, networks utilizing technologies such as intelligent mesh or control plane are having tangible impact on the successful global flow of information, and their significance is bound to grow in the years to come.
These networks’ performance – and specifically their resilience and the ability to self-heal – allows them to address technology challenges (e.g. network downtime) before they become business challenges (e.g. inability to execute a transaction).
From the end customer perspective, they truly are working faster than real time.