Files
ai-xray/config_test_results.md
2026-03-08 07:09:56 +00:00

45 lines
2.2 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Xray Configuration Comparison — Port 443
**Date**: 2026-02-20 15:49:15
**Local xray**: 25.10.15 | **Remote xray**: 26.2.6
**Remote**: share.alogins.net (83.99.190.32), LXD container "xray"
## Results
| Configuration | Status | Avg Latency | P95 Latency | Download | Upload |
|---------------|--------|-------------|-------------|----------|--------|
| A: TCP+Reality+Vision (baseline) | ✓ OK | 301ms | 1225ms | 26.78 Mbps | 2.99 Mbps |
| B: XHTTP+Reality | ✓ OK | 142ms | 225ms | 27.54 Mbps | 3.44 Mbps |
| C: gRPC+Reality | ✓ OK | 144ms | 250ms | 18.83 Mbps | 3.19 Mbps |
| D: TCP+Reality+Vision+Fragment+Noise | ✓ OK | 393ms | 452ms | 36.10 Mbps | 3.18 Mbps |
## Configuration Descriptions
| Config | Transport | Port | Flow | DPI Target |
|--------|-----------|------|------|------------|
| A | TCP + Reality | 443 | xtls-rprx-vision | Baseline — standard Russia anti-DPI |
| B | XHTTP + Reality | 443 | none | Volume-based TCP freezing (split uploads, XMUX padding) |
| C | gRPC + Reality | 443 | none | H2 pattern analysis (looks like enterprise API traffic) |
| D | TCP + Reality + Fragment/Noise | 443 | xtls-rprx-vision | TLS ClientHello DPI signature (fragment chains + noise) |
## Winner: B — XHTTP + Reality
**XHTTP is now the active configuration** (remote inbound + local outbound `juris-xhttp`).
- Best average latency: **142ms** (vs 301ms for baseline TCP+Vision)
- Best P95 latency: **225ms** (vs 1225ms for baseline — 5× improvement)
- Best upload: **3.44 Mbps**
- Download comparable to baseline: **27.54 Mbps**
Config A (`juris-reality`, TCP+Vision) remains as fallback outbound on local server.
### Why XHTTP beats TCP+Vision here
TCP+Vision (baseline) showed a severe P95 spike (1225ms) — likely caused by the volume-based TCP freezing that TSPU applies. XHTTP splits uploads into multiple small HTTP POST requests, preventing the single-connection volume trigger. XMUX multiplexing also reduces per-request overhead.
## Methodology
- Each config tested sequentially on the **same port 443**
- Remote inbound swapped on-the-fly via 3x-ui API
- 10 latency samples to google.com
- 10 MB download from Cloudflare speed test
- 5 MB upload to httpbin.org