Persistent IPv6 Issues and Lack of Reliability of CF

Dear Cloudflare Community and Support Team,

I am compelled to bring a critical issue from our organization to this public forum after private support tickets did not lead to any help.

Short intro about us:
First of all, we are a provider of SaaS software in the healthcare sector within the DACH market. We supply a Practice Management System that our customers entirely rely on to deliver vital services to their patients. The nature of our work demands high standards of reliability and performance, which is why we chose to use Cloudflare services.

Our latest recent issues & debugging efforts:
Over the past two days, we have received many reports from customers that our SaaS platform is slow or not loading, rendering it practically unusable for those affected. Initially, we noticed that only customers from ISP provider Telekom were affected and thought maybe there was an issue related to the ISP. However, we couldn’t validate that, and it might just be a coincidence since they have a huge market share in Germany. After some hours of debugging on customer-affected devices, we found out that the root cause is related to IPv6. While we noticed massive packet loss and long RTT via IPv6, everything was fine via IPv4. Disabling IPv6 (in CF or the routers of customers) mitigates the issue, suggesting a specific problem with IPv6 handling, particularly concerning CF’s services, as other services not relying on CF still worked without any issue via IPv6. Moreover, we’ve come across numerous reports within the Cloudflare community and beyond, indicating that many others are experiencing similar challenges. While we cannot confirm if these are directly related to our situation, sadly, there is again no clear and transparent communication from CF on these reports.
Our temporary solution — disabling IPv6 compatibility in CF settings — only partially addresses the problem and also seems not to extend to CF-hosted content (like CF Pages). Also advising our customers to disable IPv6 is neither practical nor acceptable as a permanent fix, especially since we can’t control the network environment of our customers. What is particularly alarming is the random and unpredictable nature of these issues, which have surfaced out of the blue and seem to be randomly gone for now. But each time they happen, we are directly affected due to the deep integration of CF services into ours. So, we question now: why did this even happen, and what’s the root cause/solution for it?

An additional issue is compounded by the fact that over the past years, every significant long downtime and disruption of our service has been directly linked to CF issues. Many of them are publicly well-known, while others seem to be hushed up. We engaged CF’s services to enhance the reliability and speed of our platform, but the repeated problems have led us to question whether our partnership with CF is more detrimental than beneficial. We really want to like CF, but they make it hard… As you can read, it stems from a place of honest frustration and disappointment with CF services and transparent support.

We require a detailed explanation of the root cause and a swift resolution to this recent surge in IPv6-related problems. Did anyone else experience similar?
We hope a public discussion can lead to a long-term solution, not only for us but for all customers experiencing similar issues, now and in the future.

Thanks and regards,
Johannes