How SURF reduced the indirect traffic of a CDN from 58% to 20%

About SURF

SURF is the national research and education network (NREN) of The Netherlands. As a collaborative organization, SURF works with members of Dutch education and research institutions to ensure and increase the quality, efficiency and security of their nationwide networks.

Challenges

SURF had held a direct interconnect with a well-renowned CDN since 2019. At the end of 2023 the IP Transit providers of SURF changed, and since then, SURF had been receiving 48% of the CDN traffic indirectly through one of the new transit providers. Without BENOCS Analytics, they didn’t have the visibility of the direct vs. indirect traffic, so this was an undiscovered challenge – or rather, a missed chance.

Before: SURF traffic from March to May 2024

Solution

BENOCS Analytics offers network operators full visualization off all traffic traversing the network, Having seen the capabilities it offers, the team at SURF decided to invest in the tool to functionally and financially optimize their network. Specifically the ability to see direct and indirect traffic was an integral part of the decision for the peering team: a Sankey diagram that displays the traffic flow, as well as a time series and statistical table that reveal traffic levels over time, would uncover various opportunities to optimize their peering strategy.

90%
increase in direct traffic
100%
traffic visibility

Implementation

After Flow Analytics was deployed, BENOCS’ data quality assurance team noticed the large amount of indirect traffic from a major CDN and informed Joachim, network engineer at SURF. Joachim was already in contact with the BGP Engineer of the CDN in question, and suggested corrections to the BGP announcements to steer traffic towards the direct interconnection and not via a transit provider.

Results and Benefits

After this change was implemented by the CDN end of May 2024, the direct traffic jumped from 42% to 80%, an absolute increase of 90%. Consequently, the indirect traffic fell, from 58% to 20%, bringing the traffic flow to the IP upstream provider below the Committed Data Rate (CDR). This shift in traffic to the direct interconnection with the CDN network saved substantial monthly costs for SURF.

After: SURF traffic from Jun 2024 to Jul 2024
“BENOCS Analytics is indispensable for SURF to be able to analyze and guard peering traffic that runs over many separate interconnections (both direct and via internet exchanges). This case study is a great example in which BENOCS Analytics helped us to analyze the suboptimal flow of traffic between a CDN network and SURF, in this case resulting in saved costs for IP transit. In general, the Border Planner module provides a regular usage health check on peering interconnections, as well as during events (e.g. during DDoS, traffic shifts), when it is a great solution for analyzing traffic flows; especially for traffic that originates from a network with which there is no direct peering.”
Niels den Otter
Teamlead Network Experts
SURF

Future Plans

In the coming months, SURF will cooperate with BENOCS to implement BGP-LS functionality within the product. For SURF this provides a secure and scalable way to update the network topology within BENOCS automatically, and be able to visualize backbone traffic.

Furthermore, SURF will work with BENOCS to look at traffic in multiple VRFs and analyze traffic flows in separate services that run on the network.

Conclusion

Implementing BENOCS Analytics was a lightbulb moment for SURF: upon seeing the amount of indirect traffic entering their network, the Dutch NREN was able to adjust their peering accordingly. This resulted in substantial financial savings and optimized traffic flow. BENOCS Analytics is now an essential tool in SURF’s peering management and strategy.

Back to news