WhatsApp Server Serves Over One Million TCP Connections

WhatsApp · FreeBSD · Erlang · kqueue

alizar September 23, 2011 at 16:21

WhatsApp Server Serves Over One Million TCP Connections

The developers of the popular WhatsApp Messenger program reported on the blog that they had optimized the server backend to improve performance, uptime and scalability. As a result, they managed to achieve that one server serves more than a million tcp sessions. WhatsApp backend runs on FreeBSD + Erlang. They are not the first to achieve a million simultaneous tcp sessions on a single server.

$ netstat -an | grep -c EST

1016313

Tags:

Tags

#WhatsApp #FreeBSD #Erlang #kqueue

Read Next

BPMN in 1C: process recording automation

BPMN in 1C: process recording automation

Learn how the tool captures actions in 1C and generates BPMN diagrams in 12 minutes instead of 3 days. Save 80% of time on modeling. For 1C analysts.

AI agents for QA: 11 pipeline skills

AI agents for QA: 11 pipeline skills

Multi-agent system automates 80% QA: from requirements decomposition to autotests in GitLab. Architecture, skills, integrations for middle/senior QA. Study the implementation on Claude Code Skills.

LLM and originality: experiment with apophatic AI | Analysis

LLM and originality: experiment with apophatic AI | Analysis

Experiment with Gemini 3.1 Pro proved: LLM reproduces original concepts without citing the source. How to check the uniqueness of AI generation? Find out now.

TeleClaude for Claude Code in Telegram

TeleClaude for Claude Code in Telegram

Self-hosted TeleClaude bot replaces OpenClaw: isolated Claude Code sessions by Telegram topics. Installation, memory, compression. Deploy locally in 5 minutes.

Fortress Loop in IT: protection vs development

Fortress Loop in IT: protection vs development

Developers spend resources on endless protection, losing growth. How to break the loop: risk balance, universal measures and focus on incomes. For middle/senior specialists.

AWS Braket Direct: analysis of pricing and risks of the quantum service

AWS Braket Direct: analysis of pricing and risks of the quantum service

Analyzing the launch of AWS Braket Direct: real task price from $75, hidden limitations of the 200-qubit processor, and forecast for the quantum industry. Find out who benefits.

Safe Module Refactoring: Preserve Behavior | IT Practices

Safe Module Refactoring: Preserve Behavior | IT Practices

How to Extract Modules from a Monolith Without Regressions. Boring Refactoring Strategy with Feature Flags and Tests. Learn What Not to Change in the First Stage.

Putin's Technological Alliances: Russia's New Strategy

Putin's Technological Alliances: Russia's New Strategy

Analysis of Putin's Statement on International Technological Alliances: What Lies Behind the Shift in Course, Who Benefits, and How the Dual-Circuit System Changes the Sanctions Landscape.

Trend 'Turn Messages into a Song': Neural Networks Turn Fights into Hits

Trend 'Turn Messages into a Song': Neural Networks Turn Fights into Hits

Neural networks turn breakup chats into pop ballads. How Suno and TikTok make millions from others' pain, ethics, and forecasts. Read.

AGI CPU Arm: 272 cores for AI data centers

AGI CPU Arm: 272 cores for AI data centers

Arm AGI CPU on Neoverse V3 offers 272 cores per server and over 45,000 in a rack. Integration with Altera FPGA for AI. Details of characteristics and configurations — read in the review.