In February, I proposed and led a panel session at APRICOT 2025, discussing the merits and ability of Internet Exchange Points (IXPs) to contribute to measuring the resilience of the Internet.
My premise for this discussion was twofold:
- IXPs’ unique vantage points can provide local and regional level availability and performance insights that public international measurement projects cannot measure as accurately.
- The data that IXPs can collect and share with their members and the Internet measurement and development communities can be used as a value-added service to help grow and sustain IXPs.
To validate these thoughts, I was joined by:
- Philip Smith, who has been helping establish IXPs globally for 20+ years and assisting with the RouteViews project.
- Christoph Dietzel, who is the global head of products and research at DE-CIX, one of the largest global IXP companies operating IXPs in more than 40 locations.
- Anand Raje, who helped establish and manage Kolkata IX in India.
- Amreesh Phokeer, my Internet Society colleague who has a background in Internet measurement research and has been instrumental in developing the Pulse IXP Tracker.
Below are some key themes and outcomes from the discussion and the recording.
What perspective can IXPs offer?
The Internet has become increasingly dense and complex as it’s evolved. More than 68,000 advertised networks currently connect trillions of devices across every economy, sea, and ocean, as well as those travelling through the skies and space.
Some measurement projects monitor the performance of these networks using physical or software ‘probes’ and ‘servers’ located along routes or at endpoints like the modem in your house. Other projects, like RouteViews, monitor and archive live changes to the global routing system, logged in the Border Gateway Protocol (BGP) table, which determines the paths that data packets travel across the Internet.
The resolution of the data these projects collect is proportional to the number of unique vantage points they can establish. As Philip noted, “One operator’s view of the [BGP] table may be completely different from somebody else.”
The challenge that Anand clarified is that not everyone wants to share this information, a reason why he and others established India’s Indigenous Internet measurement system (AIORI-IMN). IXPs offer researchers one avenue that overcomes this challenge to a certain degree.
IXPs offer a centralized vantage point within local Internet ecosystems where many networks connect. They can monitor for and measure:
- Traffic patterns, including traffic volumes and flow dynamics across the multiple networks that connect to them. Spikes, drops, or anomalies can indicate outages, congestion, or routing issues affecting availability to Internet users in the regions they service.
- Route diversity and redundancy between networks. More peering routes at an IXP usually suggest better availability for users.
- Sudden changes in traffic or route announcements, which often precede or indicate large-scale Distributed Denial-of-Service (DDoS) attacks, route hijacks associated with Border Gateway Protocol (BGP) anomalies, and Internet outages.
- Malicious activity, such as botnet communications or scanning.
What would an IXP measurement system look like?
Many IXPs provide standardized data plane and control metrics to their members and the public. Many also register their details in public interconnection databases, such as PeeringDB, from which the Pulse IXP Tracker draws data.
The data plane handles the movement and processing of data packets across a network.
The control plane manages the network’s configuration and routing policies.
One way to think of it is that the data plane is the ‘driver’ and the control plane is the ‘GPS’ that directs the driver.
Christoph suggested that many researchers would also be interested in measuring the management plane, as it would help explain the business decisions behind irregular traffic flows. However, he doesn’t envision many networks being willing to share their ‘secrets’.
Christoph also clarified that DECIX and other IXPs are not sharing or analysing all the raw data they collect due to privacy requirements and computational limitations — trying to crunch such large sums of data would take decades. Instead, DECIX researchers are sampling the data to answer “fundamental questions [about] how the ecosystem changes, grows, or evolves… for the good of the Internet.”
Amreesh talked about the importance of using and supporting a diversity of measurement systems to help validate the collected data. He offered the example of how Routing Information Service and Packet Clearing House provide a similar service to RouteViews and are often used in tandem to validate each other.
A comment from the audience was the need for data to be understandable, something that Christoph said DECIX has grappled with over the years. “We’ve gotten better over the past years, but most of the [data] we [share] is not consumable by my mother, which is the right abstraction for most of the people in this [economy] and around the world.”
Another audience member suggested developing a standardized method to show the economic impact of peering at IXPs. Amreesh noted that the Internet Society is currently developing this feature and hopes to make it available on the Pulse website soon.
Is there interest from IXPs in doing more measurement?
In a word, yes. Most of our IXP operator-dominated audience recognized the value of the data they could collect and contribute. The limiting factor, though, is how to resource it.
Philip appealed to the commercial organizations that have benefitted and continue to benefit from free measurement services for support. “Please return some of the proceeds of what you have done to help the volunteers and the exchange points and everywhere else who are [sharing] all this information, who are putting all this infrastructure in place usually out of the goodness of [their] heart.”
Anand noted that the Indian government has been a significant champion in understanding the availability and resilience of the Internet in India. So has the local academic community, with several universities agreeing to host probes and the large sums of data they are collecting in return for having access to the data for their own comparable research.
Christoph added how academic institutes can attract and cultivate the next generation of IXP operators and advocates.
The Internet Society’s Sustainable Peering Infrastructure Funding Program also offers financial and technical support to enhance existing IXPs. Applications will be accepted from 28 January to 10 April 2025 and from 1 July to 21 August 2025.
Ultimately, I’m hoping this is the start of the conversation. With greater interest in the IXP and measurement community, we can work towards a standard solution that enables all IXPs to contribute to improving our understanding and ultimately strengthening the resilience of the Internet.
Robbie Mitchell is a Senior Communication and Technology Advisor at the Internet Society, with interests in Internet measurement and resilience.
Originally published on the Internet Society’s Pulse Blog.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.