Server Churn Insights: Balancing Efficiency and Accuracy in OpenVPN Probing

Table of Links

Abstract and 1 Introduction

2 Background & Related Work

3 Challenges in Real-world VPN Detection

4 Adversary Model and Deployment

5 Ethics, Privacy, and Responsible Disclosure

6 Identifying Fingerprintable Features and 6.1 Opcode-based Fingerprinting

6.2 ACK-based Fingerprinting

6.3 Active Server Fingerprinting

6.4 Constructing Filters and Probers

7 Fine-tuning for Deployment and 7.1 ACK Fingerprint Thresholds

7.2 Choice of Observation Window N

7.3 Effects of Packet Loss

7.4 Server Churn for Asynchronous Probing

7.5 Probe UDP and Obfuscated OpenVPN Servers

8 Real-world Deployment Setup

9 Evaluation & Findings and 9.1 Results for control VPN flows

9.2 Results for all flows

10 Discussion and Mitigations

11 Conclusion

12 Acknowledgement and References

Appendix

7.4 Server Churn for Asynchronous Probing

After the Filter generates a list of probing targets, the Prober can either send probes synchronously as soon as a target is emitted, or asynchronously, waiting for a pre-configured interval before sending probes to targets in batches. Sending probes synchronously has the advantage of obtaining the most accurate results before the server IP is churned. However, this requires the probing system to be online the whole time. In contrast, sending probes in batches is more efficient and easier to manage, but the server IP may be churned if the interval between the filtering phase and the probing phase is excessively long. We explore a probing frequency that achieves efficiency and accounts for possible server IP churn. To do this, we monitor the 180,858 known OpenVPN servers from the Censys Set described in 6.3. Starting from August 2nd, 18:00 EDT, we probe the servers every 3 hours for a week and record their responses.

As shown in Figure 9, even after a week, only 2.39% of OpenVPN servers either are not in the listening state or have been replaced by a different service. This suggests that the majority of OpenVPN servers are not churned frequently. In

Figure 9: OpenVPN server churns over time.

our online evaluation, we choose to probe targets in batches on a daily basis to balance between efficiency and potential IP churn. Based on the result of this test, approximately 0.9% of servers may be churned within 24 hours.