Ericsson, the top global supplier of mobile network gear, recently created a new unit focused on network analytics and control. SDxCentral’s Matt Palmer sat down with Jan Haglund, VP and Head of Network Analytics and Control at Ericsson, to talk about how today’s evolving networks are changing the role of those functions.
SDxCentral: What is the primary mission for the network analytics and control product area?
Haglund: Three areas of focus summarize our mission going forward: consolidation, automation, and exposure. First, the consolidation of IP networks and the whole movement to SDN is a very important step because consolidating network resources is a prerequisite for so many other things. Bringing together the control system in a much more centralized architecture is what gives you programmability, feature speed, feature development, and agility. In the traditional telco space, consolidation supports a big movement called user data centralization. One of the things we’re driving with the new telco-grade database technology is control and analytics systems around that.
Another piece of consolidation is around network management. The fragmentation of network management is a huge hindrance today in our industry because it consumes a lot of people, drives up cost, and gets in the way developing new systems. We’re developing solutions to bring resources together.
The next step is very much about automating processes because automation frees up resources, takes out cost, and speeds things up. Rather than having someone figure out when to extend capacity or change parameters, the system can decide that and do it based on a set of predefined rules. That can be applied both to the radio networks space as well as transport and cloud networks. And it can be applied to more than networks. If as a subscriber end up here in Sweden and your home operator figures out that you’d be better served with another set of services, you could actually automate your whole marketing campaign around that.
The final step, exposing capabilities in networks, is going to become more important going forward. Many operators are limited in what they know about how the network is performing. Figuring out if the customer experience is good enough is a challenge by itself — now a lot of companies want to make sure they’re getting what they’re paying for. Having a handle on the system will provide this information and enable things like providing new services on the fly.
We’re definitely seeing database consolidation on the SDN side. We’ve observed this is one of the biggest inhibitors to enable things like group policy. What are you doing with cloud and IP guys to help solve that?
Haglund: Now we touch the security aspects of things. You don’t want to let just anyone into your network. I think the same thing translates into the cloud space. You only want certified and approved applications to run in your cloud. Making sure you have authentication handling of that is going to be so much more important to us, especially with network function virtualization.
An even bigger trend is that of non-SIM devices coming on to networks. How do you make sure that other devices that are not authenticated by the traditional mechanisms of mobile systems are authenticated in an equally strong way? That will require its own set of databases, handling, among other requirements.
How will user data management help bridge the gap between the traditional 3G, 4G, and future 5G, with Wi-Fi?
Haglund: I think the way for anyone to be successful with wireless technology is to leverage the radio networks they already have. It’s all about coordinating your resources and making sure that users have the best experience.
In the future there will be a need for additional heterogeneous networks like small-cell development, which is really the combination of wide-scale coverage and smaller-scale coverage. It’s important that you treat the whole thing as one coordinated system, whether it’s for capacity or coverage. You can call that SDN, you can call it radio network control – it’s just different names for the same thing. A Wi-Fi network will never replace the macro coverage built up by an LTE network. It’s simple math.
I like how you related radio network control to SDN, because a lot of what SDN does is what you have been doing on the radio side for the last 10 to 15 years.
Haglund: Yes. Whether you’re responsible for LTE or a virtual cloud network, you’re looking for the same things around policy and control.
What’s your view of automation and orchestration in terms of partnerships and third-party products?
Haglund: The whole idea of SDN is that you introduce a wider variety of ecosystems for interoperability – both northbound and southbound. Anyone can gain some speed by doing something proprietary. But the fact that you can use the same mobile phone and roam worldwide from California – that would not have worked without interoperability. That’s why we have stayed true to our decision to support the open source initiatives for SDN. We’re working with other companies southbound to make sure we can open up interfaces as well.
Which southbound interfaces do you expect to win out over time?
Haglund: I think we should avoid any religious debate around that. The most important thing is not which particular protocol, but the fact that there is openness and interoperability. I think everyone benefits from that in the long run because that’s how to get the potential of SDN technology to penetrate the market. If we only work on silos, it’s not going to take off at the same pace.
Now that Ericsson’s product areas have been rejiggered, our audience wants to know what specific technologies and products are within the analytics and control group.
Haglund: Real-time analytics with consolidated databases is the fundamental starting point for everything. So that’s one key piece where we’re building not only expertise but also products and solutions. Management of hybrid networks is the starting point for all SDN and cloud-based deployment. Second is policy control. We have to go beyond the standard PCRF mechanisms that exist today. PCRF creates a policy for subscribers, people like you and me, on how the network should react when we try to do certain things But policy will be important in many other ways as well. For example, determining what do you do when a VNF runs out of resources. The group also covers SDN technologies, which are all about adding programmability to different domains of the network: the transport side, the data center, and the services domain, as with service chaining.
On the analytics side, is Ericsson participating in any open source databases?
Haglund: Yes, that’s going to be very important for us. Open source will be important for interoperability between different vendors. Having said that, the carrier-grade and reliability of this is so, so important. You read about incidents where someone has had problems with user databases and they took down the whole network. You can’t imagine the effort we’ve spent in Ericsson to make sure we have stability at the database and system level.
What do you think are the initial key uses cases to make network analytics real today? Security? Marketing? Price? Optimization?
Haglund: I think it’s all of the above. Investment is another. You want to make sure you have the right reliability and flexibility. Otherwise it triggers a churn or a revenue loss in your network because a lot of the branding in our industry is based on reliability.
Marketing is a good one, too. The business used to be pretty predictable. People were using voice services and all marketing had to figure out was how many minutes of voice you’re actually using. That time is way past, so now we’re into data. But what’s actually going on over that data network? Application coverage is so important to figure out, and it’s always changing.
How about security?
Haglund: Security used to just be about denial of service attacks and whatnot. But there are other types of network anomalies you need to pay attention to, such as anomalies that could occur via devices coming in by mistake, or, God forbid, to take down parts of the network. You’re going to see some cool solutions from Ericsson around that going forward. Also, as cloud technology penetrates into the carrier space, security protection and analytics are important to achieve the reliability and security requirements you have in that space.
If you had to summarize, who are the top three customers by use case or type for network analytics?
Haglund: Today it’s definitely the big carriers. They’re trying to figure out if the network is performing as it should, if the users are getting the experience they’re paying for. If I look a little bit ahead, you’re going to see other industries becoming more dependent on communications and, in particular, mobile infrastructure. They’re going to put demands on analytics because they want to figure out what’s going on.
What do you see as the role of SDN in terms of tying together this notion of consolidation, automation, and exposure?
Haglund: If you look at all these issues, SDN is the way to get it done. If you see that you need to reconfigure your network, or want your VM to be somewhere else for latency or capacity reasons, or need to add a service to a certain category of users – an SDN-based application where your networking resources are represented in software is the way to get it done. Otherwise you have to more or less physically go out and rewire your network, and everyone knows that that takes time that we don’t have.
SDN is becoming the DevOps platform for networking. The technology automatically enables you to stitch together an ecosystem of different vendors products and open source technologies into custom offerings that expose new network capabilities.
How will NFV affect analytics and control use cases?
But with all those benefits come challenges. You could argue that some things in your network become less predictable. If something fails in a service chain of distributed functions, it becomes less obvious where to fix it and how to fix it. If you move some of your real-time critical applications on top of the cloud, things like latency and real-time properties may become less predictable also. This makes the analytics piece even more important.
It seems the traditional tools you would use for latency are also no longer useful.
Haglund: Yes, but it’s all solvable. It’s also about capacity detection and flexibility. For example, when natural disasters happen in different places of the world, you have huge traffic increases over just a few minutes. How do you respond to that in a flexible way and on that kind of time scale? It requires a lot of detection mechanisms, and I think that’s why SDN and NFV and analytics probably are part of this sort of holy triangle of tomorrow’s networks.
What’s really neat about this conversation compared to how it would be from a year ago is that it sounds like we’re having a conversation with the head of a software business instead of a networking hardware business.
Haglund: That’s a good reflection. Obviously a lot of value is being created in software, but hardware is going to continue to be important for sure. We are evolving with our customers.
We covered a lot of interesting ground. Thanks for your time.
Haglund: Thank you.
For more on Ericsson’s efforts to redefine network control and analytics, make sure you read this announcement from Mobile World Congress 2015