After rolling out a bandwidth-on-demand MPLS-based service, BT is working on launching another flavor of bandwidth on demand.
Jim Sabey, head of BT Networking and compute sales specialists, spoke to FierceTelecom about BT's new bandwidth-on-demand internet service, his company's work with the TM Forum and how BT is using automation and machine learning during this Q&A.
As its name implies, bandwidth on demand gives BT's business customers the flexibility to increase their bandwidth as needed to meet peak needs. Bandwidth on demand could also lower the cost structure for businesses by giving them more control of their networks. Bandwidth on demand is a feature on BT IP Connect Global, which is BT’s wide area networking service available in 198 countries and territories.
Orange Business Services, AT&T and Verizon have trialled or deployed their own versions of bandwidth-on-demand services using universal customer premises equipment (uCPE) or virtualized customer premises equipment (vCPU). UCPE and vCPU allow service providers to offer virtualized services to their business customers on lower cost x86 servers without the need for proprietary appliances.
This Q&A with Sabey was lightly edited for length and context.
FierceTelecom: BT launched its bandwidth-on-demand service in September of last year, can you give us an update on it?
Jim Sabey: Customers have rolled it out and have deployed it and they really like the flexibility of going to the BT portal and being able to address the bandwidth increases. It's been going well.
An update that I can share as well is that since we launched our IP Connect Global MPLS, we’re also very close to launching a similar bandwidth-on-demand service for our Internet Connect Global offering. Being able to offer that across the internet is going to be a huge differentiator. It will also be helpful to our customers from an internet perspective as more and more of them migrate from MPLS to more of the internet-type of solution. With our global multinational organization, they will have the ability to now flex up or down the bandwidth that’s required on internet. (Sabey said in a follow up email that the internet-on-demand service was slated to launch in September.)
Fierce Telecom: How is BT using AI, machine learning and automation?
Sabey: AI, machine learning and automation, all those are very important, especially when you get into analytics. Arguably, BT owns and operates the largest network in the world, so what are we going to do with that information and how are we going to help other customers with that as well and within similar verticals? How do we apply lessons learned and all that? AI and machine learning and automation are really important, especially with emerging technologies like SD-WAN and NFV.
We're doing a project with the TM Forum where we're actually working on standardization of component architectures and interface requirements. It's about viewing that rich telemetry, and providing the data to perform AI and machine learning on a network to help optimize it and really provide that insight on customer usage. We picked four of BT’s top customers in the U.S. We did a little bit of digging and research on them and we provided them with over 70 reports.
FierceTelecom: The key is what do with all of the analytics information once it has been collected?
Sabey: Are they actually doing anything with some of those reports? Probably not, just to be truthful and transparent. But this all ties into the project that we're doing with TM Forum. There really isn't a concrete standardization, if you will, but the purpose of the project with TM Forum and BT is to really start developing those standard for what we would call the operational AI model to enable them to understand, discover and monitor all of these things in real life.
It’s going to be really interesting to see how that continues to progress because AI and machine learning and a lot of those things are really going to define the architectural components moving forward; databases and deployments and things of that nature and how things are going to continue to progress in this field.
I have customers tell me all the time “Hey Jim, we would love to have a solution that basically learns and auto learns, and if there's an error somewhere in the network, it just automatically fixes it and then it sends me an email, so I at least know that something changed and if I needed to reverse course.”
You're seeing some of that with application aware routing, and even with our Cisco SD-WAN service, for example. We actually have a demonstration where we can show how SD-WAN actually works. You have a site that has two circuits and when some of the packet loss on one of the circuits goes up really high the traffic automatically reroutes over the other available line. Some of that is automation, but taking that next step into the AI and the machine learning, it’s going to be really neat to see and interesting to see because it really is important as SD-WAN take that next step.