tor

Tor’s Speed Bottleneck: Is Conflux the Answer?

If you’ve used Tor for anything serious—research, market access, secure communication—you already know the pain point: it’s slow. That may be changing sooner than later.

Sam Bent

Mar 10, 2025 — 3 min read

Anyone who's spent more than a few minutes on the Tor network has felt it: latency, lag, painfully slow page loads. Whether you're accessing censored information, communicating anonymously, or using it for more gray-area reasons, speed has always been Tor’s Achilles heel.

And while the Tor Project has made technical progress in areas like security and resistance to abuse, performance has largely remained stagnant—until now.

A long-standing proposal called Conflux might be the shift the network needs. It hasn’t been deployed yet, but recent developments suggest it’s finally being taken seriously. So what exactly is Conflux, and why does it matter?

The Core Problem: Tor Is Single-Threaded by Design

Tor routes traffic through a single circuit consisting of three relays: a guard (which knows your IP), a middle (which doesn’t know anything useful), and an exit (which sees where you're going, but not who you are). This structure is crucial for preserving anonymity—but it's also vulnerable to bottlenecks.

If any relay along that path is slow or congested, your entire connection suffers. Add in proof-of-work defenses introduced in response to denial-of-service attacks, and Tor becomes even slower for legitimate users.

In short: the system prioritizes anonymity, not efficiency.

What Is Conflux?

Conflux is a proposed solution to this problem, first outlined in Proposal #329. It changes the way Tor builds circuits by introducing the idea of dual-circuit routing.

Instead of routing all traffic through one fixed path, Conflux uses two separate circuits in parallel. These circuits share the same exit relay but use different guards and middles. This creates a more flexible and resilient route for data.

Conflux also actively monitors the performance of both circuits, shifting traffic between them based on congestion, latency, and throughput. In some cases, it even splits packets between circuits and reorders them on arrival to maintain proper sequencing.

This design is inspired by multi-path transport protocols but adapted for anonymity-focused networks.

Why It Matters

If implemented successfully, Conflux could:

Reduce page load times significantly by avoiding congested paths
Improve reliability by automatically failing over to the better-performing circuit
Make traffic analysis more difficult, since data is fragmented across multiple routes
Increase adoption, by making Tor more usable for mainstream users who don’t want to trade privacy for performance

However, it's important to note that Conflux is not primarily a security enhancement. Its focus is on performance, though its side effects may improve certain anonymity properties.

Implementation Status

As of now, Conflux is still a proposal, but it was recently referenced in an official Tor Project blog post discussing the Arti client—a Rust-based rewrite of Tor's core functionality. The post mentioned that groundwork for Conflux is being laid within Arti’s architecture.

This matters because Arti is modular and modern, making it far more capable of supporting experimental features like Conflux compared to the legacy C-based implementation.

The fact that the Tor Project is acknowledging Conflux publicly is a strong signal that it’s moving from theoretical to possible.

Risks and Trade-Offs

Of course, routing data over two circuits isn't trivial. The design introduces complexity:

Memory usage increases because out-of-order packets must be buffered until they can be properly reassembled.
Side-channel risks may emerge if adversaries try to correlate circuit behavior.
Relay load may spike, as relays now have to manage two circuits per client instead of one.

These aren't deal-breakers, but they are important concerns—especially for a network that already operates on constrained volunteer infrastructure.

Why This Is Worth Watching

The Tor network is critical infrastructure for journalists, whistleblowers, activists, and anyone operating in hostile environments. It's also widely used by darknet market participants and others who rely on anonymity for operational security.

If Conflux can meaningfully improve performance without compromising core anonymity guarantees, it could broaden the Tor user base and make the network far more resilient overall.

That, in turn, increases the anonymity set for everyone—something the entire privacy community benefits from.