The continuous growth of data traffic is a real challenge for operators today, as multiple services add their traffic to operator networks. In any infrastructure planning, digital twins are an important tool for saving costs and anticipating difficulties and conflicts. This also applies to the planning of networks.
Greenfield and brownfield simulations play an important role in planning & realization of smart cities and industrial development. Greenfield simulations focus on undeveloped areas, providing a baseline assessment of the potential and risks of new projects. In contrast, brownfield simulations focus on the optimization and revitalization of used or abandoned urban areas. Both approaches provide valuable insights for planning and implementing future development projects.
Now we would like to present a project with our customer EWE Tel, in which siticom has taken a leading role. In an interview with Gerhard Weber, Principal Consultant at siticom, we learn more about the challenges, as well as the customer benefits and future prospects that emerged from the project.
Hi Gerhard, could you give us a brief overview of the project with EWE Tel, focusing on simulation?
Hello, yes of course. The trigger for the simulation was the very fast increasing utilization of EWETel's network, and they expect the rate of increase to be 25% to 50%. It is clear that in the existing network the link utilization will soon exceed the limit of 40% per link. 40% link utilization should not be exceeded in normal operation, in order to transport traffic via a second link in case of failure. The remaining free percentages are intended for traffic peaks.
The simulation was set up in order to have the possibility to try out different solutions. Different solutions here refers mainly to different network topologies (strict, light or non-hierarchical) and the locations of traffic sources (caches). Caches are servers on which movies are stored, for example. For example, there are caches of Netflix or Amazon.
Where are the focal points?
Mainly, the simulation is used to study the network utilization of different topologies with their traffic sources. For this purpose, traffic data is necessary. This data is provided by EWE Tel. For example, it is known how much load has been brought into the network by e.g. Netflix since the year 2013. From this, corresponding forecasts for the coming years can be calculated and thus also simulated. A good 20 different load types with individual growth rates are included in the simulation.
In the meantime, the simulation has been expanded again and again. It is now also possible to
fail routers and connections,
make statements about the propagation delay of packets,
to simulate PPPoE subscribers,
operate the network with 1,000,000 routes, and
to simulate the L3VPN and EVPN of the business customers in accordance with the real network.
What are the customer benefits here?
By using a GUI based on OpenStreetMap, network topologies can be created and simulated very easily. Also existing networks can be examined in this way. Existing networks can be optimized or rebuilt by migration steps that are also "tested" in the simulation.
The greatest advantage is probably to increase the security of network planning and thus to avoid weaknesses in one's own network in advance. This can be realized very cost-effectively with the help of simulation.
Without simulation, it only becomes apparent in reality whether decisions once made were correct. Mistakes here can be very expensive.
What is the outlook for the future? What would be further possibilities?
Currently, only Juniper routers (vMX) are used in the simulation. We are currently working on opening the simulation to other routers. This will then also allow interop and feature "tests". For example, the simulation can be used to test whether the routers of other manufacturers can also handle 1 million routes, or whether L3VPNs work between routers from different manufacturers, or how they have to be configured to work.
Flexibilization of traffic sources is also being considered. Right now, no routers are allowed to fail in the simulation that have a traffic source connected to them. We have already discussed various solutions for this, but have not yet decided on one.
In principle, it would also be possible to reproduce the connections in the simulation on the basis of GIS data. This would allow better statements to be made about running times. In addition, it might also be possible to calculate line costs in more detail.
Thank you Gerhard for this insight. If you want more information about greenfield and brownfield simulation, please have a look at our Insight.