Traffic Optimization

Who

Companies: CDNs, CSPs

Roles: Traffic engineers, peering managers, mapping experts

Situation

A CDN is experiencing sluggish content delivery despite the short geographical distance their content needs to travel. CDNs not only need to regionalize servers to solve localization challenges but they also need to localize users to regionalize the traffic and by that avoid issues affecting quality of experience. These challenges are currently mainly overcome by mapping traffic based on internal or third-party Geo-IP data, which neglects the actual topology and routing of the networks. A solution is needed to make a calculated decision based on the operator’s network topology and route traffic based on the actual network distance, capacity and utilization provided to you in real-time via an API.

Challenge

A CDN provider has to feed traffic to a couple of IP Prefixes in Heilbronn (HLB), which is located between their regional servers in Frankfurt (FRA) and Munich (MUC). The Geo-IP-based mapping algorithm decides to feed the traffic from Frankfurt as the geo distance between FRA-HLB is shortest with 234kms, but is still experiencing latency issues.

Solution

  • The peering manager of the subscriber network operator creates a customer portal in BENOCS Analytics which provides visibility of traffic exchange between the operator and CDN in a 4-6 dimensional view.
  • Based on the operator’s network topology, the traffic from Frankfurt is routed to Stuttgart (STU) first and then to Heilbronn.
  • After analyzing the traffic, it becomes clear that by also taking the outbound routing into consideration, the traffic is taking the following path: FRA-STU-HLB-STU-MUC-FRA. This round trip is about 735 Kms which is more than three times the geo distance.
  • The CDN ingests data from BENOCS Flow Director which allows a metric data exchange real time via an API.
  • The metric can include hopcount, kilometer, capacity, utilization for thousands of prefixes and hundreds of routers in an operator’s network.

Results

  • Mapping decisions made based on ground truth
  • Asymmetrical traffic paths made visible
  • Terabits of misallocated traffic removed from backbone and better quality of experience for subscribers achieved
  • OpEX and CapEX savings

Takeaway

While the trend to the edge will positively influence the customer experience and potentially sink traffic distribution costs, it will also require better localization and stability. With BENOCS Flow Director you can steer your network traffic along the optimal route, speeding up content delivery time and saving on backbone costs.

Back to Solutions