What kind of Internet would we design today if we were to start from scratch using what we now know?
In 2007, attempts to answer this question at Stanford University  led to a project by networking gurus and experts to analyze the present and future of IP networks.
Amazingly enough it’s only been ten years since the first iPhone was released. To look back on that first version is a fascinating experience and makes you realize just how far smartphones have come. (more…)
In a highly dynamic technological sector led by powerful multinationals that put considerable emphasis on branding, international presence and acceptance of any kind for small and medium-sized businesses is both noteworthy and appreciated.
This week I have written the article below to be published in “ComunicacionesHoy”, a well known telecommunications magazine and online media website in Spain. As the article has only been published on ComunicacionesHoy in Spanish, Teldat has translated it into English so that all our non-Spanish speaking social media followers can also read it. I hope that you enjoy my article.
Even though, at first glance, opting for Internet access lines to build corporate networks seems like a sound decision cost-wise, the truth is this choice brings far more benefits when it comes to control, flexibility and user-friendly management.
By focusing on this much wider point of view, carriers have the opportunity of putting together a very attractive offer to ensure their clients benefit from the enormous range of possibilities this type of network has to offer.
Imagine we undertake a survey, asking CIOs of large companies with a wide network of offices or remote points (banks, insurance and travel agencies, distribution chains, etc.), what they would want to improve in their company communication network. Requests for a wider bandwidth would certainly arise, together with the possibility of selecting the network/access technology that best suits their remote needs, the use of multiple, efficient, redundant and simultaneous unfettered access, higher network intelligence to dynamically adapt to real time communications and the full atomization of operating tasks (to name but a few). Moreover, all respondents will ask to pay a fraction of the price they are currently paying (for communications based on MPLS networks) without compromising either SLA or security.
Sounds too good to be true? It isn’t, as you’ll see
The answer to CIOs’ prayers is SD-WAN, a communications architecture made up of different pieces of technology (some new, others not) that is able to produce a synchronized performance capable of satisfying the most exacting of CIO aspirations. The SD-WAN base is made up of internet lines and a further layer providing SLA said accesses don’t have, obtained from traffic engineering over several internet links or by maintaining MPLS access (using far less bandwidth for critical corporate traffic). SD-WAN is essentially made up of the following:
a) Virtual Private Network (VPN) over any IP access, MPLS or Internet, offering complete freedom to select your access technology (fiber, DSL, LTE, etc.), the highest security and without limiting the number of accesses used at remote points.
b) Traffic selection, in order to identify the applications that use the network and apply different policies (depending on the criteria of each application in relation to business).
c) Real time quality analysis of the access to remote divisions, based on traffic monitoring and usually through synthetic traffic (polling).
d) Network intelligence that makes it possible to dynamically adapt different applications over different accesses, depending on the policies defined for said applications and the state of the accesses.
e) Visibility of network behavior with respect to applications and the use of said accesses.
f) Centralized network control, which permits unified global parameterizing of behavior and automated provision of remote point elements.
Actually, much of this technology isn’t new. Applying traditional techniques you can, for example, use internet lines as an access method. Secure VPNs can then be employed to balance applications (depending on their granularity or access status), while obtaining greater visibility on network usage. However, implementing such a network using traditional methods would be a Herculean task! Given how complex it is to configure separate network elements so that they operate as a single network, just attempting it would be crazy. Here however, is where the SD of SD-WAN networks really comes into play.
SDN (Software Defined Networks) have clearly demonstrated their value in Data Centers, integrating different systems, automating both management and service chains, providing a virtual view of the network thus enabling global management from a single management point.
This same idea, applied to WAN, is what holds together the different pieces of this puzzle with apparent simplicity. While complexity still exists, it is hidden under a layer of abstraction that facilitates both the implementation and management of an SD-WAN network. Like SDN management, SD-WAN supports the simple, unified and central parameterizing of network behavior to adapt to new applications or modify existing application policies.
In this context, three SD-WAN product supplier groups have emerged: one being companies evolving from the connectivity sector towards SD convergence; a second, in direct contrast, proposing consolidated SDN solutions and extending these solutions towards the wide area network, and finally a third group, involving start-ups specifically focused on SD-WAN. Whatever the case however, analysis of this solution must be rigorous and should, at the very least, keep the following in mind:
a) To use open standards/protocols to ensure the network is not reduced to a single supplier.
b) Scalability for both network design and speed.
c) Capacity to cover network terminator features to unify access and SD-WAN features.
d) Appropriate traffic granularity to ensure the balance of applications (complying with business parameters) without compromising network performance.
e) Active polling to check the health status of network accesses regardless of traffic.
f) Centralized management tools for unified network design, provisioning and management.
g) Cost of network elements.
SD-WAN is still in the early stages of development, the real number of implementations being low. However it’s definitely on the radar for the majority of IT departments who are planning network migration. Significant SD-WAN growth is expected in the near future, firstly as a complement to MPLS and, in the long term, as an alternative and preferred network.