Challenges with Monitoring 4G LTE Infrastructure


Matt Goldberg discusses three key challenges when monitoring service delivery over LTE infrastructures:

1. Monitoring end-to-end LTE performance across complex architectures
2. Monitoring multi-vendor environments from a single platform
3. Delivering dashboards, reports, and alerts with speed at scale


Hi. I'm Matt Goldberg with SevOne. Today I'm gonna talk to you about a couple of challenges with monitoring LTE networks. As you can see here, we've broken it down into a couple of unique buckets. You've got the RAN, the backhaul, the packet core, the transport and the data center and of course way back inside the data center, the IMS core. If we start all the way out here at the RAN, I mean, when you start thinking about managing an infrastructure end to end, you really want to start about looking all the way out at the cell site.

You want to look at the cell site itself. Out here along with the transport that's going back. Typically, behind your cell tower you're going to have a box that that's an eNodeB. Which is going to do some of the translation of the data and signaling coming off the wireless infrastructure. Then what you'll have is an actual cell site router in today's infrastructure. What this allows is very high capacity for the carriers all the way out at the cell site. This is to enable some of the new services that you start to see come out. Such as VoLTE, certainly LTE with the through put so that it allows you to do the streaming video that a lot of us are consuming today on the wire.

Then you have your backhaul which is your standard transport. Typically you have some larger sized routers that are actually doing point to point. In some cases and certainly in some countries around the world, carriers buy their backhaul capacity from other carriers. This is a very tightly measured segment between these two orange lines. What that allows carriers to do is actually see how other carriers are performing so that you can manage and understand what is the quality that you are delivering to your consumer who has their mobile phone over here.

You then have your evolved packet core. Really what sits in here are items like your packet gateways, your authentication, your authorization systems. A lot of these systems today are based on servers. Its a lot of heavy applications that are sitting on top of very unique servers. Some of which have been purposely built, others, certainly as things have evolved today we are starting to see virtualized versions of this piece of software.

Then as you move into your data center, you're getting things like your firewalls. You'll have your firewall. You'll have your load balancers as well. You'll have your internal LAN switches. You'll also of course have more and more racks of servers that are providing some of your backend services.

Finally, all the way over here, what we have is your IMS core. This is really where, certainly, we start talking about technology such as voice over LTE and some of the other services that carriers are starting to roll out. This is really where a lot of these applications live back here. This controls the calls, your visual voice mail, streaming video, and all these reside again typically in today's environment on a very large set of servers that are geographically either clustered or geographically spaced. That carriers can route your traffic to the closest area.

One of the challenges, as you can see here, this is a pretty wide and deep architecture that these carriers are rolling out and its getting more and more complex as time evolves. One of the difficult things is managing this. How do you manage end to end in this infrastructure? What you want to do is to be able to follow the packet through the infrastructure starting as close to the consumer all the way over here out at the RAN. Back to the RAN through the backhaul to the packet core all the way through the transport and data center back to the IMS core. Obviously then back out to the customer. We typically see that the average number of hops that somethings taking now is twenty or thirty hops just to get a single authentication for a cell phone.

If you think about that, all that is happening in extremely fast time if you think about yourself making a phone call today. Really, you need to be able to understand every hop along the way, what's going on.

As we were just talking about the complexity of measuring these buckets, another complexity or challenge that you have in managing an infrastructure this diverse is certainly the number of vendors. Typically, you very rarely find a homogeneous infrastructure in terms of a hardware vendor. Its very heterogeneous. If you look out at the RAN, some of the typical vendors that you'll see out there Huawei, Ericsson, Alcatel, certainly your normal vendors like Cisco and Juniper are providing some of the routing and switching we were talking about from the cell site outer perspective. You've also certainly got the NID vendors in there as well that are providing the back end as we were talking previously, from an SLA compliance perspective for the backhaul.

In the backhaul, you've certainly got your typical routing and switching vendors. That's your Ericsson, your Alcatels, your Ciscos, your Junipers, etc. Huawei. In your packet core, you start to gt a little more outside of your normal vendors. Cisco provides some packet core, Alcatel, Ericsson, Huawei, Samsung. All these different vendors, as you can imagine, have very unique ways of extracting data from these systems. One of the key challenges is how do you get all this different data from all the different vendors. In a lot of cases, what we see is that you'll have one functionality provided by two different vendors. You could have a packet gateway from let's say Alcatel and Cisco for instance. How do you look at the two side by side when you're managing an infrastructure? To understand what the difference is between the two platforms, you really want to see the one one combined view of the two platforms. You don't want to be looking at different EMS's or NMS's from the hardware vendor to understand what's going on with a particular platform.

In the transport and data center, certainly we were talking about firewalls and load balancers. Certainly your routers and switches. You have your vendors like F5 and A10 and Citrix. You've got firewall vendors like Juniper, FortiGate and then certainly the 5's of the world as well. As you continue back to the IMS core, you've got your Gemaltos, your Alcatels, your Bridgewaters, your Nokia Siemens. Mavenir is another great player back there. The other piece in the data center to keep in mind too is there's vast and tremendous amounts of storage. You have folks like NetApp and EMC in there who are also playing.

How do you take all this data and really combine it into a single pane of glass? You can really understand as, again, this packet is flowing through the infrastructure and hitting all these different components what is going on as its traversing the infrastructure.

As we've been talking, we want to be able to manage this heterogeneous environment and also that complete end to end picture. The last piece of it is really understanding speed and scale. As this infrastructure continues to grow, and and as the technology continues to evolve, how do carriers manage this entire picture at extreme speeds so they understand in real time what's going on?

As technology has evolved, we've seen the need to collect data multiple different ways in the infrastructure. SNMP is historically what carriers have used and hardware vendors have provided. Nowadays we are looking at more APIs and EMS and NMS focus for some of these newer components and the packet core at the data center and the IMS core. What's interesting is the SevOne PAS really allows you, whether you deploy a physical or virtual state, to really be able to give you this end to end picture and plug into all these different unique data sources. As you've plug into these unique data sources, you really want to be able to run a report very, very quickly.

Let's say you're an operator and you're managing tens of thousands of devices with millions and millions of elements and you're creating billions of data points every day, and you want to say, "Okay, in my packet core what are my top two hundred most utilized IP pools?" Being able to run that report and generating it in just a few seconds is really the key to understanding and managing an extremely complex infrastructure like we've drawn up here on the board today.

Thank you for watching the white board video today as we've discussed how SevOne helps overcome the challenges in managing and monitoring an LTE and IMS network.