kdb Insights Enterprise
The Data Timehouse
kdb+ Time Series Database
PyKX Python Interoperability
Services & Support
Industry & IoT
Energy & Utilities
Healthcare & Life Sciences
KX University Partnerships
Partner with Us
Become a Partner
Find a Partner
Connect with Us
Paul Colgan is the SVP of Telco solutions at KX. He has over 19 years of experience working in the Telecoms software industry in roles spanning engineering, product management and go to market strategy. During this time the companies ranged from first day start-ups growing to be acquired to large publically quoted global entities acquiring others. The common thread was a focus on high performance, high throughput network-centric solutions to help Operators solve their toughest business challenges.
Q: The Telco industry seems to be at an exciting time with the rollout of 5G networks starting to happen.
Yes, 5G is the big talking point at the minute with lots of emphasis on the technology changes required to enable it. Automation, virtualization, white-boxes, network slicing, “network in a box”, SDN/NFV are all de rigueur at the moment when talking with anyone.
Q: What are all of these technology changes going to do for the Operators, as well as for you and me?
Well lots of things, both in the short and long term. Initially, Operators will look to drive more revenue by having a premium plan for 5G. If you look at what happened during the 4G rollout and what is happening in Korea today with 5G, the “first mover” Operator is looking for a reward for being first to market with the enhanced speeds. As with 4G through, I think this premium will then shortly disappear as other Operators in that country also offer 5G. That’s exactly what happened with EE in the UK, for example. Vodafone and others came out with 4G at the same price as 3G, and EE was forced to match.
For consumers, the majority will not go out and buy a new 5G-capable device and sign up to the premium plan – especially given the recent trend of consumers tending to not upgrade their devices as often as they did in the past because there is less differentiation between new models of phones and the older versions. Over time though, everyone will upgrade to a new phone which will be 5G-capable and they will get to enjoy the significant speed increases at the same or less cost than today. The way Operators rolled out 4G conditioned consumers to expect faster for the same (or less), so 5G will be no different in that sense.
Q: What about the longer term? It sounds like with 5G the consumer will get more speed for the same price, so why are Operators spending lots of money to upgrade their networks?
Yes, well this is where it gets interesting. Operators to date have followed the “build it and they will come” mantra. Nobody predicted that the iPhone concept would make 3G, or that video would make 4G. There are use cases around AR and VR but I think they will be very niche. For me, the mass market for these technologies is in the gaming sector where you sit at home in the living room with your VR headset. That will stay on your home broadband, so unless you use fixed wireless access to get your broadband into your home, that VR traffic will mainly affect your cable or fiber broadband providers networks.
The reality, like before, is that people just don’t know “the killer app” for 5G. However, one thing that is happening at the moment is a greater focus by the Operators on their Enterprise B2B channel. There is a massive push from business clients for private network rollouts. Some of this is driven from the Digital Transformation wave that is sweeping all business sectors globally, some from a cost perspective, some from a security perspective. If you take the Digital Transformation for the industrial sector as an example, factories all around the world are trying to connect their sensor data from their different systems. What they are finding is that you can’t run a plant by retrofitting cables everywhere. WiFi is not usually an option due to interference and security reasons. They are starting to deploy Private LTE networks instead to carry this sensor data. When this becomes 5G, they will be able to add in network slicing where you can have one slice for the factory operational data and another for local phone communications – like an old PBX but instead using your mobile phone just like you do on a regular cellular network. This private network will connect to the public core network when you need to communicate outside the private network. Similar setups are being used in hospitals, university campus, large hospitality centers, etc
Q: Beyond the network itself and the connectivity it gives what else should we expect from 5G?
Another aspect of 5G is that it is focused on virtualization and automation and I think this is where the Operators can improve their margins. If revenues remain flat, at best, long-term then Operators will need to run their businesses more efficiently if they are to cope with the continuous increase in data volumes. If they wish to roll out new services or add new capacity these will need to be automated. We see a renewed focus on analytics within the Operators to use data to inform and drive those automation steps. In the last number of years, most Operators have built large data warehouses or Hadoop clusters and put all of their operational data in there in the hope they can extract high-value information for their marketing, customer care or network planning teams. In the main, this approach has not delivered the type of value we all hoped for. Hadoop was originally built for the distribution of a web search engine and not for the storage and advanced analysis of time-series log data. The various add-ons and iterations made it complex to maintain. As you can see from the trials and tribulations of MapR, Cloudera and HortonWorks it hasn’t worked at a macro level either.
Q: What you describe sounds, in part, like Business Intelligence we have become used to. What additional capabilities are being enabled?
To achieve “zero-touch automation”, Operators are integrating their analytics stack into their workflow engines. The most interesting use cases require the analytics to be very real-time, highly accurate, granular and easy to maintain. The Hadoop clusters cannot provide this in a cost-effective, maintainable and scalable way. For example, you need to know in near real time when new capacity is required, so you have the time to spin up the VMs and containers needed without causing the customer experience to be impacted. When these network changes have been made, you need to know in real time if they have been successful or if you need to modify the increase again. Likewise, for performance and fault management you need to know, ideally in advance, that an issue is likely to occur so you can get ahead of it and prevent the issue, or start the remediation process sooner so that the duration of the fault affecting customers is reduced.
Even if you link the analytics and the various orchestrators within your workflows it still takes time to complete an automated workflow so the more real time, predictive and granular your analytics the leaner and cost-efficiently you can run your network and all general operations.
If you look at the recent past to the number of high profile Operator outages (EE and Vodafone Ireland only in the last couple of months for example), its clear that Operators are struggling with the complexity of their networks. They have legacy 2G components to phase out as they add 5G components and data volumes explode. There are not enough people to manage this flux so they need to codify the network intelligence and processes of their staff so they can manage all of this change along with the increase in data load.
The fact that our software can scale down to the lowest hardware environment of a vCPE as easily as it can scale upwards to accommodate the massive data volumes of a Tier 1 Operator’s network data gives clients great flexibility and cost-effectiveness. For us, it’s the same software and the same analytics. So it’s beneficial for both parties.
Q: You mentioned sensors in factories earlier. We hear a lot about the explosion in the number of IoT devices and 5G. How are they related?
Yes and it is true, the number of devices being connected is phenomenal. From the consumer side, we are automating our homes and connecting everything up to Siri, Alexa, etc. I read recently that 25% of Americans have a smart speaker from one of the major vendors. I have done some of this myself and being able to control your home from your phone is very useful, especially if you travel a lot.
NB-IoT and LTE-M are both integral to 5G. They provide low-power, low-cost communication for remote devices. They are part of the 4G specification but will continue into 5G and evolve into being more capable of carrying “heavier” content like short videos, snapshots, etc that are needed for some IoT use cases.
Some Operators are content to just sell SIM cards to enable the IoT ecosystems, whereas others like Verizon, for example, are buying fleet companies and trying to offer the service to all small to large logistics companies. Again, it is an example of how they are trying to capture new revenues by focusing on their Enterprise clients.
Q: Presumably there is a lot of focus on the cloud too?
There certainly is. The combination of virtualization, agility, and cost-efficiency that the cloud offers is about to completely transform the industry. What we are seeing is a move from distributed, proprietary hardware that defines networks physically, to cloud-based solutions running on standard Intel environments providing their software-defined equivalent. C-RAN paved the way by decoupling basebands from towers and enabling the multi-cell communication to give us self-optimizing networks. V-RAN is taking it a step further to fully centralized cloud-based, virtualized network functions that enable fully software-defined networks (SDNs).
It all points to networks that will become more powerful and more flexible, and as a result, more complex to manage and maintain – but only if done manually. The secret is to automate the entire process using real-time monitoring to detect issues, on-line updates to resolve them and predictive analysis to avoid them. To that end, the flexibility to make online configuration changes dovetails nicely with the cloud-enabled and real-time processing capabilities of kdb+ to identify and make those configuration changes automatically. Add machine learning and predictive analytics into the mix and we are moving to a world where network maintenance is more about software engineers in the office than field engineers on the road. It’s all a sweet spot for Kx.
It’s a great example of “digital transformation” that everyone is talking about today – but unlike, say, the printing world where the displacement of typesetters by desktop publishers took decades, it’s happening in the telco world as we speak