Network technologies *

Hello, Habr!

Let me say right away - I'm not a network engineer or an IT professional in the classic sense. My knowledge of networking is limited to a CompTIA Network+ certification and curiosity. So if network specialists see that something could have been done better, I'd be happy to hear constructive comments.

Now for the story.

A couple of weeks ago, I moved into a new apartment. The first thing I did was set up the internet - I called the guys from the provider to install the equipment, and I went to check the small cabinet in the wall where all the wires come in. Inside, I found a bunch of stuff: telephone wires, something unclear, and several coaxial cables. One came from the hallway, the second was labeled "living room," the third "office," and the fourth was unlabeled.

Neural Networks for the Little Ones

Every time you say “Thank you” to a neural network, you launch a pipeline that multiplies hundreds of matrices with billions of elements, and burn as much electricity as an LED lamp in a few seconds.

This is the first article in a short series dedicated to networks for AI/ML clusters and HPC.

In this series, we’ll touch on the principles of model operation and training, parallelization, DMA and RDMA technologies, network topologies, InfiniBand and RoCE, and we’ll also philosophize on the topic of general and specialized solutions.

In this particular article, we’ll figure out what a neural network is, how it works, how it’s trained, and most importantly, why it needs hundreds of expensive GPU cards and some kind of special network.

The refrain of today’s story: there’s no magic in neural networks—it’s just a multitude of simple operations on numbers, performed on computers with special chips. There’s no magic in how they work, nor in the infrastructure they run on.

The modern world is saturated with all kinds of information, and in our difficult times, it's important to be able to not only find it but also to save it. Many have probably noticed that on YouTube, besides the junk, cats, and other useless things (which we sometimes don't mind watching), there is a lot of useful material on a wide variety of topics. And sometimes it would be nice to save this material for the future, so as not to depend on the changing moods in the world.

In this article, I want to explain how you can download videos, audio (Part 1 of the article), playlists, and entire channels from YouTube (Part 2 of the article) without using a VPN and in pure Python. A quick disclaimer: we won't need a VPN, but we will create our own tool that will solve the "problem with outdated and worn-out equipment Google Global Cache" (you know what I mean). I think this tool will be especially relevant today, when for many Russians, YouTube barely works or doesn't work at all.

Hey you. Yes you. Do you have more routers than you can reasonably explain? Do you flash custom firmware for fun? Have you ever spent an entire evening troubleshooting an issue with router, then called it “relaxing”?

Xeovo is looking for you and your router, network and technical knowledge. We are going to give an opportunity for our community members to expand guides for routers, earn money and get recognition on Hub.

Our router tutorial section has the biggest gaps. Let's change this together.

Abstract. This work examines the physical foundations of Ekahau Sidekick measurements and the device offset mechanism from the perspectives of antenna theory, receiver noise theory, statistical signal theory, and the IEEE 802.11 standard family. It is shown that the scalar received signal strength indicator (RSSI) offset constitutes a linear level shift and does not model the true signal-to-noise ratio (SNR) of the client device, the quadrature amplitude modulation (QAM) constellation structure, the rate adaptation algorithm, or roaming behaviour. In addition to five independent physical and systemic sources of inaccuracy, the paper addresses modeling assumptions in Ekahau with respect to multiple-input multiple-output (MIMO) gain, multipath propagation, airtime estimation, and SNR visualisation. Verified numerical error estimates for representative deployment scenarios and practical recommendations are provided.

This is the pilot episode of our new interview series Meet the Developer, where we talk to the people behind anti-censorship tools. Our goal is to shine a light on the developers whose open-source solutions help millions of people stay connected.

In this first episode, we sit down with Toby, the lead developer of Hysteria, to discuss the project’s origins, technical challenges and his perspective on internet censorship.

Let’s start with an introduction. Tell us who you are and what you do.

Just call me Toby. I’m a software engineer. Previously, I have worked for a large company. But right now, I’m a co-founder of a startup with some friends.

Nice to meet you Toby! Would you like to share what type of startup it is, or is it a top secret project?

We are still in stealth mode.

Why did you decide to develop Hysteria?

It was originally a project I developed for myself when I was in college.
China’s global Internet connectivity has been notoriously bad for as long as I can remember (still not any better right now). Not just in the sense of censorship, but also in terms of connection quality.

For example, if you have a server in the US and want to connect to it from China, expect over 10-20% or more packet loss.

So if you set up a proxy server in another country to circumvent censorship, it would be painfully slow (the most popular tools back then were GoAgent and later Shadowsocks).

So Hysteria began as an attempt to improve my speed for watching YouTube videos.

It’s always great to see developers building something to solve their own challenges. I can relate to the packet loss issue. Either you suffer the packet loss, or you have to purchase an expensive server with CN2 routing, which will cost a lot.

This article is not about Puppeteer being a bad tool. Puppeteer is excellent. And competent TLS fingerprinting will bypass most defenses. But there is a class of tasks where even a perfect network stack won't save you — because detection has long since landed at the level of rendering engine behavior. Let's take a look at how Cloudflare and Akamai expose you through WebGL and Canvas, and why “clean” code no longer works.

Not so long ago, about a year ago, I finally received my long-awaited package. Flipper Zero. For those who don't know, Flipper Zero is a small device that allows you to engage in all sorts of fun activities in life. The fun starts with the Sub-1 GHz antenna and continues with micro-scripts for PC. If you're interested, please read on.

Yesterday I wrote about the graphical shell Launcher for GoodbyeDPI, which allows you to intuitively use the GoodbyeDPI solution from ValdikSS to solve the problem of YouTube throttling and the unavailability of a number of sites in Russia. Everything was fine, but these were solutions only for Windows. In the comments, the main questions were about what to do with Android, Linux, and Mac, and why not in the source code. Alas, the repressions of the RKN (Roskomnadzor) force the Habr administration to censor articles, I am forbidden from making changes to yesterday's article, and the link itself is only available outside of Russia, so I am explaining about Android, Linux, and Mac here, with links to the source code.

So, for Android there are a lot of projects, I liked Release ByeDPI 1.0.0 · dovecoteescapee/ByeDPIAndroid · GitHub. For Mac and Linux I would install https://github.com/xvzc/SpoofDPI/releases. All in source code.

ByeDPI for Android is an application that runs a local VPN service to bypass DPI (Deep Packet Inspection) and censorship. A SOCKS5 proxy ByeDPI is launched locally on your device and all traffic is redirected through it.

In this guide, we will install the v2rayA package on OpenWRT using the stable 23.05.0. A router with at least 128 MB of RAM (256 is preferable) and more than 16 MB of storage is recommended (the installation takes about 30 MB of storage)

v2rayA is a simple-to-use and powerful client focused on Linux. Despite its name, the current version uses xray-core, although it's also possible to use v2ray-core. It has a web interface for managing settings and importing configurations and subscriptions. It supports everything that xray-core supports:

Shadowsocks (incl. 2022), ShadowsocksR, Trojan, Vless (including XTLS-Reality, XHTTP), Vmess, Juicity, Tuic

The guide will include:

1. Installation from the repository

2. Configuring v2rayA and bypassing blocks using Re:filter, Antifilter GeoIP, Geosite

Wireshark is a widely used tool for capturing and analyzing network traffic, actively used for both educational purposes and for troubleshooting computer or network issues. Wireshark works with almost all OSI model protocols, has a user-friendly interface, and a convenient data filtering system. In addition, the program is cross-platform and supports the following operating systems: Windows, Linux, Mac OS X, Solaris, FreeBSD, NetBSD, OpenBSD.

Recently, I needed to run a VLESS subscription on Windows, make it work with my work VPN, and deal with all the associated 'joys.' I tried v2rayN, Nekoray, and Hiddify—and quickly realized I wanted to write my own client.
And so, singbox-launcher was born:
👉 https://github.com/Leadaxe/singbox-launcher

Below are a few details about why and what came of it.

Hello everyone, today we'll talk about Android TV. Bypassing the throttling on a TV seemed to be the most problematic and difficult for me. Since I have several TVs at home deprived of YouTube, and even more are waiting to be set up for friends and acquaintances, I decided to collect all the bypass methods I could find.

Since I'm a fan of self-hosting, I have a home infrastructure:

Orange Pi - a media server;

Synology - a file dump;

Neptune 4 - a 3D printer with a web interface and a camera feed. And I'd like to have secure access to it externally via my phone and PC, while also having internet access outside the RF. I used to use OpenVPN for these needs, but it's no longer reliable. So I started studying the documentation for an excellent tool from our Chinese comrades - Xray!

What you'll need:

A server with an external IP for the infrastructure. In my case, it's an Orange Pi, hereinafter - Bridge

The server you want to access - Server

A server outside the RF for internet access. Hereinafter - Proxy

A client of your choice. Hereinafter - Client

Client and server on Linux - Xray-core, which can be installed via the official Xray installation script

Client for Android - v2rayNG

More clients can be found in the Xray-core repository

Let's take the VLESS-TCP-XTLS-Vision-REALITY configuration file as a base and start reading the Xray documentation

Routing is done on the client. For example, if the client accesses the xray.com domain, we route the traffic to the Bridge, and for all other connections - to the Proxy. Then the Bridge routes the traffic to the Server if the client accessed server.xray.com.
It looks like this:

This weekend was all about 'Connection Reset.' While news channels vaguely reported that 'users are complaining about outages,' we were in chats and on test servers trying to understand the physics of the process.

We were asked to talk about the protocol technology XHTTP in the context of XRay, VLESS, and others. You asked for it, so here it is!

First, a bit of history. The classic use of VLESS and similar proxy protocols (including with XTLS-Reality) involves the client connecting directly to a proxy server running on some VPS. However, in many countries (including Russia), entire subnets of popular hosting providers have started to be blocked (or throttled), and in other countries, censors have begun to monitor connections to 'single' addresses with high traffic volumes. Therefore, for a long time, ideas of connecting to proxy servers through CDNs (Content Delivery Networks) have been considered and tested. Most often, the websocket transport was used for this, but this option has two major drawbacks: it has one characteristic feature (I won't specify it here to not make the RKN's job easier), and secondly, the number of CDNs that support websocket proxying is not that large, and it would be desirable to be able to proxy through those that do not.

Therefore, first in the well-known Tor project for bridges, the meek transport was invented, which allowed data to be transmitted using numerous HTTP request-response pairs, thus allowing connections to bridges (proxies) through any CDN. A little later, the same transport was implemented in the briefly resurrected V2Ray. But meek has two very significant drawbacks that stem from its operating principle: the speed is very low (in fact, we have half-duplex transmission and huge overhead from constant requests-responses), and due to the huge number of GET/POST requests every second, free CDNs can quickly kick us out, and paid ones can present a hefty bill.

Lately, there has been a flood of comments that goodbyedpi is not working again, so I decided to make instructions for you on 4 working ways to restore goodbyedpi's functionality. It works differently for everyone, so test them out to see which one suits you. Write in the comments what helped you, maybe some of your own values!

In a changing network infrastructure, mobile internet users face questions: what resources remain available, and what does this look like on a technical level? This material is the result of a practical study using standard network analysis tools.

No speculation—only measurements, numbers, and technical facts.

Today, the major release v2rayN v7.0 was released, and along with it, my series of commits that add support for the "Russia" preset.

For it to work, a Russian source of geo files for v2ray/sing-box/etc. was also created.