Enterprise communications player Colt is building out its SDN and NFV Novitas platform to increase its on-demand services offerings as well as broaden their reach to more customers.
Hybrid multi-cloud architecture is in demand as distributed enterprise applications grow and are increasingly utilized by a mobile and remote workforce. The geographically dispersed branches of global enterprises and the demands of the Internet of Things (IoT) stretch the capacity of single clouds.
Applications such as unified communications have boomed, which increases the complexity of managing networks as well as putting increased demand on security and performance. Concurrently, SD-WAN is in favor as hardware-based WAN management becomes hopelessly inadequate to cope with the variability in the traffic and application usage on multi-cloud networks.
Telco Transformation recently spoke with Fahim Sabir, director of architecture and development for Colt Technology Services Group Ltd . Sabir is responsible for Colt's delivery of next-generation connectivity service platforms, specializing in SDN and NFV technologies.
Telco Transformation: Colt's Novitas platform has been described as network services self-provisioned, orchestrated and consumed in near real time through automation as well as being integrated with third-party networks, including cloud networks. Presumably, this will take a leap in OSS/BSS systems and their replacement by policy control. What is your view on the state of maturity of policy control in networks? What are the holdups in achieving the goals needed for on-demand network services?
Fahim Sabir: While near real-time control and consumption of networks can benefit from policy control, it is not essential to enable them. Colt is implementing near real-time control in its Ethernet-on-demand connectivity solutions by actively configuring edge and gateway devices, predominantly via vendor-specific APIs. (See Coppens: Ethernet On-Demand Key to Colt's SDN Plans.) The following goals are sought to be achieved in the process:
FS: The development of standard APIs is still at a relatively early stage, with entities such as MEF, OASIS, and the TM Forum driving such standardization with groups of like-minded service providers, but focused in specific areas such as Metro Ethernet with MEF. For providers, this will reduce both costs and the time required to integrate with other service providers, and for Colt, this can help to extend the reach of our on-demand services across the world.
There is broad industry recognition of the need for agility and the fact that if a standard exists but has some flaws, it is more useful than having no standard at all. This may sound trivial, but it is key to rapidly iterating towards something more robust. In terms of vendor adoption of standards, this is very important. There are still a number of areas where component standards exist, such as YANG or TOSCA, but the standards to bind these together are missing.
TT: Your recent network upgrades expanded data center distributed computing, including data center-to-data center connections. How do these compare with alternatives such as fog computing, CDN networks, Apache Mesos, and a variety of edge solutions?
FS: Other connectivity solutions cannot provide the performance, security or reliability that can be provided by private connectivity solutions, such as those offered by Colt, and which are underpinned by the high-performance Colt IQ network. While not important in consumer "Internet-centric" scenarios, such qualities are key to enterprises that need to guarantee quality of connectivity between their data centers to successfully run their businesses. Colt's IQ Network is data center-centric, offering direct access to over 700 data centers across Europe, Asia and North America. Colt's network spans over 200 connected cities in nearly 30 countries, with over 24,500 on-net buildings and growing.
Organizations are increasingly relying on remotely delivered services such as cloud computing, disaster recovery, business continuity planning, content delivery, on-demand applications among many others, and as such, require pre-provisioned connectivity that links these data centers.
Service providers that offer solutions such as CDNs, cloud services or edge solutions to their customers underpin their internal infrastructure with private connectivity which gives them high performance and reliability.
TT: Your company has been trying to increase the adoption of enterprise clouds with innovations in SDN APIs to work across networks. How is this expected to increase enterprise cloud adoption? What kinds of companies and traffic flows are receptive to using such services and what pain points are addressed?
FS: The key advantages available with enterprise clouds are agility and programmability of infrastructure environments. The network that connects to these enterprise clouds is often comparatively static and underpinned by manual processes. This results in significant fixed costs for connectivity which is "sized to peak" rather than "continuously right-sized." SDN APIs for on-demand networks bring the same level of programmability to the network as is offered by cloud infrastructure, allowing "continuous right-sizing." Very few organizations require a consistent amount of capacity, so almost any organization can benefit from the flexibility offered. Other use cases include those with highly seasonal or event-based demand.
TT: How do you ensure failures are minimized when you seek to scale in a multi-site, distributed environment?
FS: Failures are ultimately unavoidable and as such, any distributed environment needs to be built with sufficient redundancy in order to minimize the impact of failures. Intelligent policy-based automatic re-route capabilities, available with solutions like Colt's SD-WAN, which is built into our Novitas platform, can redirect important traffic when specific metrics head towards failure and are available to our customers to access online to define these policies as well as see how the policies are being enacted.
Policy control ensures the efficient utilization to get the maximum cost-benefit of different types of connectivity. However, at some point the scarcity of factors outside of those controlled by policy begins to become the bottleneck, a key example here being bandwidth. Standard APIs enable the control of these factors, and when combined with policy-based control provide a powerful platform for highly efficient, and very cost-effective, connectivity.
— Kishore Jethanandani, Contributing Writer, Telco Transformation