A new routing strategy and additional relay capacity have been rolled out for Torzon users in Europe. Live measurements show a noticeable drop in latency and fewer timeouts during peak hours compared to the previous network layout.

The upgrade focuses on exit-node proximity, smarter circuit selection and mirror health weighting, so EU traffic is automatically steered toward the most responsive Torzon endpoints without manual mirror switching.

Impact Snapshot

-32%
Avg Latency
-41%
Timeouts
+27%
Session Stability
3
Regional Hubs

Latency Before vs After

Before
420 ms (EU avg)
Now
285 ms (EU avg)

Average page load times for EU users decreased from around 2.1 seconds to around 1.4 seconds on first hit, with follow‑up requests benefiting from connection reuse and caching.

Routing Strategy Changes

Mirror Health Weighting

Mirrors are now scored continuously based on response time, error rate and recent uptime. Clients are guided toward mirrors with the best current score instead of a static list.

Regional Hubs

Traffic from EU exit nodes prefers EU‑adjacent mirrors when they are healthy, reducing the number of long trans‑continental circuits and improving consistency.

Fast Failover

If a mirror starts timing out or returns repeated errors, it is temporarily deprioritized and traffic is shifted to healthy endpoints within a few seconds.

// simplified scoring
score(mirror) = f(latency, uptime_24h, error_rate);

if (score(primary) < threshold) {
  primary = select_best_mirror(region = "EU");
}

Current Mirror Hubs

EU Hub (Frankfurt)
Healthy · Latency ~260 ms
US Hub (New York)
Healthy · Latency ~340 ms
Fallback Cluster
Healthy · Used only on failover

Regional hubs are abstracted behind the official onion entry points. Users do not need to manually choose a mirror; the network layer handles routing decisions.

Back to All Posts