We want to inform our customers about an ongoing incident with one of our transit providers, GTT Communications, which is currently not causing a negative impact on our network. In accordance with our communication standards, we decided to still publish it as a precautionary measure.
At 05:19, our NOC was alerted that one of our 400G ports with GTT was down at our EQX-FR4 location. During diagnosis, we found the issue to be on GTT's side and created a ticket with their NOC. Right after creation, at 05:51, a second 400G port went offline. Neither outage caused any visible disruption according to our real-time network overview, as convergence happens almost instantly if a router port goes down physically.
At 06:36, the GTT NOC claimed that the issue was outside of their network and advised us to check our cross-connects with the datacenter operator (Equinix).
At 06:56, the two remaining 400G ports we have with GTT at EQX-FR4 went down, causing about 15 seconds of packet loss on a few routes because the ports failed simultaneously and GTT traffic shifted automatically to our second PoP at DRT-FRA8. We escalated the matter to GTT immediately, but did not receive feedback for several hours.
At 11:21, we were informed that the reason for the outage is an ongoing cooling system failure at EQX-FR5, where two water leaks had been detected and isolated.
Please be aware that this does not affect our EQX PoP directly, as the FR4 datacenter is unaffected. Temperatures were already checked manually overnight and found to be optimal.
For the time being, the situation is as follows:
- Our GTT ports at DRT-FRA8 are still working. We always operate our transit ports at both PoPs. Other transit providers (Tata, NTT) are unaffected.
- Our transit capacity is reduced from 4.8 Tbps to 3.2 Tbps, which is still more than sufficient even for large-scale DDoS attacks. Also, multiple Tbps of peering capacity are entirely unaffected. However, depending on the distribution of individual attacks, short-term bottlenecks are possible on single links.
- As traffic to and from GTT is routed via DRT-FRA8, we are experiencing an (expected) imbalance on internal backbone links between our datacenters. We are closely monitoring all load parameters and will take action if needed.
- As various network operators are affected by this issue, customers might see slightly increased latencies. This is not necessarily an issue within our control, but our support team is happy to check if you notice anything unusual.
Once the GTT ports recover, traffic will automatically shift back to our EQX-FR4 PoP. This can cause brief reconnects for existing TCP and UDP sessions on IP addresses with active DDoS protection.
Posted Aug 15, 2026 - 13:00 CEST