daarion-admin/microdao-daarion
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
fb268ec0e2a28a488ef730a33ec045daa9dbdaa9
microdao-daarion/site/greenfood/scaling_to_1000_agents/index.html
Apple ef3473db21 snapshot: NODE1 production state 2026-02-09 
Complete snapshot of /opt/microdao-daarion/ from NODE1 (144.76.224.179).
This represents the actual running production code that has diverged
significantly from the previous main branch.

Key changes from old main:
- Gateway (http_api.py): expanded from ~40KB to 164KB with full agent support
- Router: new /v1/agents/{id}/infer endpoint with vision + DeepSeek routing
- Behavior Policy: SOWA v2.2 (3-level: FULL/ACK/SILENT)
- Agent Registry: config/agent_registry.yml as single source of truth
- 13 agents configured (was 3)
- Memory service integration
- CrewAI teams and roles

Excluded from snapshot: venv/, .env, data/, backups, .tgz archives

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-02-09 08:46:46 -08:00

1329 lines
56 KiB
HTML
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.
<!doctype html>
<html lang="en" class="no-js">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width,initial-scale=1">
<link rel="canonical" href="https://IvanTytar.github.io/microdao-daarion/greenfood/scaling_to_1000_agents/">
<link rel="icon" href="../../assets/images/favicon.png">
<meta name="generator" content="mkdocs-1.5.3, mkdocs-material-9.5.18">
<title>Масштабування GREENFOOD до 1000+ агентів - DAARION Documentation</title>
<link rel="stylesheet" href="../../assets/stylesheets/main.66ac8b77.min.css">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Roboto:300,300i,400,400i,700,700i%7CRoboto+Mono:400,400i,700,700i&display=fallback">
<style>:root{--md-text-font:"Roboto";--md-code-font:"Roboto Mono"}</style>
<script>__md_scope=new URL("../..",location),__md_hash=e=>[...e].reduce((e,_)=>(e<<5)-e+_.charCodeAt(0),0),__md_get=(e,_=localStorage,t=__md_scope)=>JSON.parse(_.getItem(t.pathname+"."+e)),__md_set=(e,_,t=localStorage,a=__md_scope)=>{try{t.setItem(a.pathname+"."+e,JSON.stringify(_))}catch(e){}}</script>
</head>
<body dir="ltr">
<input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
<input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
<label class="md-overlay" for="__drawer"></label>
<div data-md-component="skip">
<a href="#greenfood-1000" class="md-skip">
Skip to content
</a>
</div>
<div data-md-component="announce">
</div>
<header class="md-header md-header--shadow" data-md-component="header">
<nav class="md-header__inner md-grid" aria-label="Header">
<a href="../.." title="DAARION Documentation" class="md-header__button md-logo" aria-label="DAARION Documentation" data-md-component="logo">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 0 0 3-3 3 3 0 0 0-3-3 3 3 0 0 0-3 3 3 3 0 0 0 3 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54Z"/></svg>
</a>
<label class="md-header__button md-icon" for="__drawer">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M3 6h18v2H3V6m0 5h18v2H3v-2m0 5h18v2H3v-2Z"/></svg>
</label>
<div class="md-header__title" data-md-component="header-title">
<div class="md-header__ellipsis">
<div class="md-header__topic">
<span class="md-ellipsis">
DAARION Documentation
</span>
</div>
<div class="md-header__topic" data-md-component="header-topic">
<span class="md-ellipsis">
Масштабування GREENFOOD до 1000+ агентів
</span>
</div>
</div>
</div>
<script>var media,input,key,value,palette=__md_get("__palette");if(palette&&palette.color){"(prefers-color-scheme)"===palette.color.media&&(media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']"),palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent"));for([key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
<label class="md-header__button md-icon" for="__search">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0 1 16 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.516 6.516 0 0 1 9.5 16 6.5 6.5 0 0 1 3 9.5 6.5 6.5 0 0 1 9.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5Z"/></svg>
</label>
<div class="md-search" data-md-component="search" role="dialog">
<label class="md-search__overlay" for="__search"></label>
<div class="md-search__inner" role="search">
<form class="md-search__form" name="search">
<input type="text" class="md-search__input" name="query" aria-label="Search" placeholder="Search" autocapitalize="off" autocorrect="off" autocomplete="off" spellcheck="false" data-md-component="search-query" required>
<label class="md-search__icon md-icon" for="__search">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0 1 16 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.516 6.516 0 0 1 9.5 16 6.5 6.5 0 0 1 3 9.5 6.5 6.5 0 0 1 9.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5Z"/></svg>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M20 11v2H8l5.5 5.5-1.42 1.42L4.16 12l7.92-7.92L13.5 5.5 8 11h12Z"/></svg>
</label>
<nav class="md-search__options" aria-label="Search">
<button type="reset" class="md-search__icon md-icon" title="Clear" aria-label="Clear" tabindex="-1">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 6.41 17.59 5 12 10.59 6.41 5 5 6.41 10.59 12 5 17.59 6.41 19 12 13.41 17.59 19 19 17.59 13.41 12 19 6.41Z"/></svg>
</button>
</nav>
</form>
<div class="md-search__output">
<div class="md-search__scrollwrap" data-md-scrollfix>
<div class="md-search-result" data-md-component="search-result">
<div class="md-search-result__meta">
Initializing search
</div>
<ol class="md-search-result__list" role="presentation"></ol>
</div>
</div>
</div>
</div>
</div>
</nav>
</header>
<div class="md-container" data-md-component="container">
<main class="md-main" data-md-component="main">
<div class="md-main__inner md-grid">
<div class="md-sidebar md-sidebar--primary" data-md-component="sidebar" data-md-type="navigation" >
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--primary" aria-label="Navigation" data-md-level="0">
<label class="md-nav__title" for="__drawer">
<a href="../.." title="DAARION Documentation" class="md-nav__button md-logo" aria-label="DAARION Documentation" data-md-component="logo">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 0 0 3-3 3 3 0 0 0-3-3 3 3 0 0 0-3 3 3 3 0 0 0 3 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54Z"/></svg>
</a>
DAARION Documentation
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../public/" class="md-nav__link">
<span class="md-ellipsis">
Home
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../public/getting-started/" class="md-nav__link">
<span class="md-ellipsis">
Getting Started
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../public/architecture-overview/" class="md-nav__link">
<span class="md-ellipsis">
Architecture
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../public/daiS_daos_overview/" class="md-nav__link">
<span class="md-ellipsis">
DAIS & DAOS
</span>
</a>
</li>
<li class="md-nav__item md-nav__item--section md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
<label class="md-nav__link" for="__nav_5" id="__nav_5_label" tabindex="">
<span class="md-ellipsis">
Internal
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="1" aria-labelledby="__nav_5_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_5">
<span class="md-nav__icon md-icon"></span>
Internal
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5_1" >
<label class="md-nav__link" for="__nav_5_1" id="__nav_5_1_label" tabindex="0">
<span class="md-ellipsis">
Infra
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="2" aria-labelledby="__nav_5_1_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_5_1">
<span class="md-nav__icon md-icon"></span>
Infra
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../internal/infra/INFRA_AUTOMATION_PACK_V1/" class="md-nav__link">
<span class="md-ellipsis">
Infra Automation Pack v1
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../internal/infra/monitoring_overview/" class="md-nav__link">
<span class="md-ellipsis">
Monitoring Overview
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../internal/infra/nodes_registry_v0/" class="md-nav__link">
<span class="md-ellipsis">
Nodes Registry v0
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5_2" >
<label class="md-nav__link" for="__nav_5_2" id="__nav_5_2_label" tabindex="0">
<span class="md-ellipsis">
Specs
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="2" aria-labelledby="__nav_5_2_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_5_2">
<span class="md-nav__icon md-icon"></span>
Specs
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../internal/specs/matrix_presence_aggregator/" class="md-nav__link">
<span class="md-ellipsis">
Matrix Presence Aggregator
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../internal/specs/city_map_spec/" class="md-nav__link">
<span class="md-ellipsis">
City Map Spec
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../internal/specs/node_join_protocol_draft/" class="md-nav__link">
<span class="md-ellipsis">
Node Join Protocol (Draft)
</span>
</a>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-sidebar md-sidebar--secondary" data-md-component="sidebar" data-md-type="toc" >
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--secondary" aria-label="Table of contents">
<label class="md-nav__title" for="__toc">
<span class="md-nav__icon md-icon"></span>
Table of contents
</label>
<ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
<li class="md-nav__item">
<a href="#_1" class="md-nav__link">
<span class="md-ellipsis">
🎯 Архітектурні виклики
</span>
</a>
<nav class="md-nav" aria-label="🎯 Архітектурні виклики">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#1" class="md-nav__link">
<span class="md-ellipsis">
1. Паралелізм та черги
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#2-state-management" class="md-nav__link">
<span class="md-ellipsis">
2. Державне управління (State Management)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#3-llm" class="md-nav__link">
<span class="md-ellipsis">
3. LLM інфраструктура
</span>
</a>
<nav class="md-nav" aria-label="3. LLM інфраструктура">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#a-llm-horizontal-scaling" class="md-nav__link">
<span class="md-ellipsis">
A. Розподілені LLM (Horizontal Scaling)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#b-llm" class="md-nav__link">
<span class="md-ellipsis">
B. Використання хмарних LLM для пікових навантажень
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#c-llm" class="md-nav__link">
<span class="md-ellipsis">
C. Кешування відповідей LLM
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#4-observability" class="md-nav__link">
<span class="md-ellipsis">
4. Моніторинг та observability
</span>
</a>
<nav class="md-nav" aria-label="4. Моніторинг та observability">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#a-prometheus-grafana" class="md-nav__link">
<span class="md-ellipsis">
A. Метрики (Prometheus + Grafana)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#b-distributed-tracing-jaeger-tempo" class="md-nav__link">
<span class="md-ellipsis">
B. Distributed Tracing (Jaeger / Tempo)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#c-elk-stack-loki" class="md-nav__link">
<span class="md-ellipsis">
C. Логування (ELK Stack / Loki)
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#5" class="md-nav__link">
<span class="md-ellipsis">
5. Ізоляція та безпека
</span>
</a>
<nav class="md-nav" aria-label="5. Ізоляція та безпека">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#a-circuit-breaker-pattern" class="md-nav__link">
<span class="md-ellipsis">
A. Circuit Breaker Pattern
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#b-sandbox" class="md-nav__link">
<span class="md-ellipsis">
B. Sandbox для кожного агента
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#6-api" class="md-nav__link">
<span class="md-ellipsis">
6. Дані та API
</span>
</a>
<nav class="md-nav" aria-label="6. Дані та API">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#a-connection-pooling" class="md-nav__link">
<span class="md-ellipsis">
A. Connection Pooling
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#b-read-replicas" class="md-nav__link">
<span class="md-ellipsis">
B. Read Replicas
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#c-api" class="md-nav__link">
<span class="md-ellipsis">
C. Кешування на рівні API
</span>
</a>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#_2" class="md-nav__link">
<span class="md-ellipsis">
📊 Орієнтовні потужності
</span>
</a>
<nav class="md-nav" aria-label="📊 Орієнтовні потужності">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#13" class="md-nav__link">
<span class="md-ellipsis">
Поточна архітектура (13 агентів)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#1000" class="md-nav__link">
<span class="md-ellipsis">
Для 1000 агентів
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#_3" class="md-nav__link">
<span class="md-ellipsis">
Необхідна інфраструктура
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#_4" class="md-nav__link">
<span class="md-ellipsis">
Орієнтовний бюджет
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#_5" class="md-nav__link">
<span class="md-ellipsis">
🛠️ План масштабування
</span>
</a>
<nav class="md-nav" aria-label="🛠️ План масштабування">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#phase-1-proof-of-concept-current" class="md-nav__link">
<span class="md-ellipsis">
Phase 1: Proof of Concept (Current)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#phase-2-production-ready-1-3" class="md-nav__link">
<span class="md-ellipsis">
Phase 2: Production Ready (1-3 місяці)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#phase-3-scale-out-3-6" class="md-nav__link">
<span class="md-ellipsis">
Phase 3: Scale Out (3-6 місяців)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#phase-4-enterprise-scale-6-12" class="md-nav__link">
<span class="md-ellipsis">
Phase 4: Enterprise Scale (6-12 місяців)
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#_6" class="md-nav__link">
<span class="md-ellipsis">
💡 Швидкий старт для масштабування
</span>
</a>
<nav class="md-nav" aria-label="💡 Швидкий старт для масштабування">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#1_1" class="md-nav__link">
<span class="md-ellipsis">
1. Додай черги (швидко)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#2" class="md-nav__link">
<span class="md-ellipsis">
2. Додай моніторинг (швидко)
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#3-load-balancer-llm" class="md-nav__link">
<span class="md-ellipsis">
3. Додай load balancer для LLM (середньо)
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#_7" class="md-nav__link">
<span class="md-ellipsis">
🎯 Висновок
</span>
</a>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-content" data-md-component="content">
<article class="md-content__inner md-typeset">
<h1 id="greenfood-1000">Масштабування GREENFOOD до 1000+ агентів<a class="headerlink" href="#greenfood-1000" title="Permanent link">&para;</a></h1>
<p><strong>Поточний стан</strong>: 13 агентів, 4 crews<br />
<strong>Мета</strong>: 1000+ паралельних агентів для сотень комітентів</p>
<hr />
<h2 id="_1">🎯 Архітектурні виклики<a class="headerlink" href="#_1" title="Permanent link">&para;</a></h2>
<h3 id="1">1. <strong>Паралелізм та черги</strong><a class="headerlink" href="#1" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: 1000 агентів = 1000+ одночасних LLM-викликів<br />
<strong>Рішення</strong>:
- <strong>Черги завдань</strong>: Redis Queue (RQ), Celery, або NATS JetStream
- <strong>Rate limiting</strong>: Обмеження кількості одночасних запитів до LLM
- <strong>Пріоритизація</strong>: VIP-клієнти → звичайні → фонові завдання</p>
<div class="codehilite"><pre><span></span><code><span class="c1"># Приклад з Celery</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">celery</span><span class="w"> </span><span class="kn">import</span> <span class="n">Celery</span>
<span class="n">app</span> <span class="o">=</span> <span class="n">Celery</span><span class="p">(</span><span class="s1">&#39;greenfood&#39;</span><span class="p">,</span> <span class="n">broker</span><span class="o">=</span><span class="s1">&#39;redis://localhost:6379&#39;</span><span class="p">)</span>
<span class="nd">@app</span><span class="o">.</span><span class="n">task</span><span class="p">(</span><span class="n">priority</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="c1"># Високий пріоритет</span>
<span class="k">def</span><span class="w"> </span><span class="nf">process_urgent_order</span><span class="p">(</span><span class="n">order_id</span><span class="p">):</span>
<span class="n">crew</span> <span class="o">=</span> <span class="n">GREENFOOD_CREWS</span><span class="p">[</span><span class="s2">&quot;fulfill_order&quot;</span><span class="p">]</span>
<span class="k">return</span> <span class="n">crew</span><span class="o">.</span><span class="n">kickoff</span><span class="p">()</span>
<span class="nd">@app</span><span class="o">.</span><span class="n">task</span><span class="p">(</span><span class="n">priority</span><span class="o">=</span><span class="mi">9</span><span class="p">)</span> <span class="c1"># Низький пріоритет</span>
<span class="k">def</span><span class="w"> </span><span class="nf">generate_analytics_report</span><span class="p">():</span>
<span class="n">crew</span> <span class="o">=</span> <span class="n">GREENFOOD_CREWS</span><span class="p">[</span><span class="s2">&quot;monthly_settlement&quot;</span><span class="p">]</span>
<span class="k">return</span> <span class="n">crew</span><span class="o">.</span><span class="n">kickoff</span><span class="p">()</span>
</code></pre></div>
<hr />
<h3 id="2-state-management">2. <strong>Державне управління (State Management)</strong><a class="headerlink" href="#2-state-management" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: Кожен агент має контекст, пам'ять, стан діалогу<br />
<strong>Рішення</strong>:
- <strong>Redis</strong> для швидкого кешу (поточні діалоги, сесії)
- <strong>PostgreSQL</strong> для персистентного стану (історія, транзакції)
- <strong>Memory Service</strong> (вже є в проєкті!)</p>
<div class="codehilite"><pre><span></span><code><span class="c1"># Приклад збереження стану</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">redis</span><span class="w"> </span><span class="kn">import</span> <span class="n">Redis</span>
<span class="kn">import</span><span class="w"> </span><span class="nn">json</span>
<span class="n">redis_client</span> <span class="o">=</span> <span class="n">Redis</span><span class="p">(</span><span class="n">host</span><span class="o">=</span><span class="s1">&#39;localhost&#39;</span><span class="p">,</span> <span class="n">port</span><span class="o">=</span><span class="mi">6379</span><span class="p">,</span> <span class="n">db</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">save_agent_state</span><span class="p">(</span><span class="n">agent_id</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">state</span><span class="p">:</span> <span class="nb">dict</span><span class="p">):</span>
<span class="n">redis_client</span><span class="o">.</span><span class="n">setex</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;agent:</span><span class="si">{</span><span class="n">agent_id</span><span class="si">}</span><span class="s2">:state&quot;</span><span class="p">,</span>
<span class="mi">3600</span><span class="p">,</span> <span class="c1"># TTL 1 година</span>
<span class="n">json</span><span class="o">.</span><span class="n">dumps</span><span class="p">(</span><span class="n">state</span><span class="p">)</span>
<span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_agent_state</span><span class="p">(</span><span class="n">agent_id</span><span class="p">:</span> <span class="nb">str</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">dict</span><span class="p">:</span>
<span class="n">state</span> <span class="o">=</span> <span class="n">redis_client</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;agent:</span><span class="si">{</span><span class="n">agent_id</span><span class="si">}</span><span class="s2">:state&quot;</span><span class="p">)</span>
<span class="k">return</span> <span class="n">json</span><span class="o">.</span><span class="n">loads</span><span class="p">(</span><span class="n">state</span><span class="p">)</span> <span class="k">if</span> <span class="n">state</span> <span class="k">else</span> <span class="p">{}</span>
</code></pre></div>
<hr />
<h3 id="3-llm">3. <strong>LLM інфраструктура</strong><a class="headerlink" href="#3-llm" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: Ollama на одній машині не витримає 1000 запитів/хв<br />
<strong>Рішення</strong>:</p>
<h4 id="a-llm-horizontal-scaling">A. Розподілені LLM (Horizontal Scaling)<a class="headerlink" href="#a-llm-horizontal-scaling" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="c1"># docker-compose для LLM кластера</span>
<span class="nt">services</span><span class="p">:</span>
<span class="w"> </span><span class="nt">ollama-1</span><span class="p">:</span>
<span class="w"> </span><span class="nt">image</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">ollama/ollama</span>
<span class="w"> </span><span class="nt">environment</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">OLLAMA_NUM_PARALLEL=4</span>
<span class="w"> </span><span class="nt">ollama-2</span><span class="p">:</span>
<span class="w"> </span><span class="nt">image</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">ollama/ollama</span>
<span class="w"> </span><span class="nt">environment</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">OLLAMA_NUM_PARALLEL=4</span>
<span class="w"> </span><span class="nt">ollama-3</span><span class="p">:</span>
<span class="w"> </span><span class="nt">image</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">ollama/ollama</span>
<span class="w"> </span><span class="nt">environment</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">OLLAMA_NUM_PARALLEL=4</span>
<span class="w"> </span><span class="nt">llm-load-balancer</span><span class="p">:</span>
<span class="w"> </span><span class="nt">image</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">nginx:alpine</span>
<span class="w"> </span><span class="c1"># Балансування навантаження між ollama-1, 2, 3</span>
</code></pre></div>
<h4 id="b-llm">B. Використання хмарних LLM для пікових навантажень<a class="headerlink" href="#b-llm" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="c1"># Гібридна стратегія</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_llm_provider</span><span class="p">(</span><span class="n">priority</span><span class="p">:</span> <span class="nb">str</span><span class="p">):</span>
<span class="k">if</span> <span class="n">priority</span> <span class="o">==</span> <span class="s2">&quot;urgent&quot;</span><span class="p">:</span>
<span class="k">return</span> <span class="s2">&quot;cloud_deepseek&quot;</span> <span class="c1"># Швидкий хмарний LLM</span>
<span class="k">elif</span> <span class="n">is_local_available</span><span class="p">():</span>
<span class="k">return</span> <span class="s2">&quot;local_qwen3_8b&quot;</span> <span class="c1"># Локальний Ollama</span>
<span class="k">else</span><span class="p">:</span>
<span class="k">return</span> <span class="s2">&quot;cloud_deepseek&quot;</span> <span class="c1"># Fallback на хмару</span>
</code></pre></div>
<h4 id="c-llm">C. Кешування відповідей LLM<a class="headerlink" href="#c-llm" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">functools</span><span class="w"> </span><span class="kn">import</span> <span class="n">lru_cache</span>
<span class="kn">import</span><span class="w"> </span><span class="nn">hashlib</span>
<span class="nd">@lru_cache</span><span class="p">(</span><span class="n">maxsize</span><span class="o">=</span><span class="mi">10000</span><span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_llm_response_cached</span><span class="p">(</span><span class="n">prompt</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">agent_id</span><span class="p">:</span> <span class="nb">str</span><span class="p">):</span>
<span class="c1"># Кеш для повторюваних запитів</span>
<span class="n">cache_key</span> <span class="o">=</span> <span class="n">hashlib</span><span class="o">.</span><span class="n">md5</span><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;</span><span class="si">{</span><span class="n">agent_id</span><span class="si">}</span><span class="s2">:</span><span class="si">{</span><span class="n">prompt</span><span class="si">}</span><span class="s2">&quot;</span><span class="o">.</span><span class="n">encode</span><span class="p">())</span><span class="o">.</span><span class="n">hexdigest</span><span class="p">()</span>
<span class="n">cached</span> <span class="o">=</span> <span class="n">redis_client</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;llm_cache:</span><span class="si">{</span><span class="n">cache_key</span><span class="si">}</span><span class="s2">&quot;</span><span class="p">)</span>
<span class="k">if</span> <span class="n">cached</span><span class="p">:</span>
<span class="k">return</span> <span class="n">cached</span>
<span class="n">response</span> <span class="o">=</span> <span class="n">call_llm</span><span class="p">(</span><span class="n">prompt</span><span class="p">,</span> <span class="n">agent_id</span><span class="p">)</span>
<span class="n">redis_client</span><span class="o">.</span><span class="n">setex</span><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;llm_cache:</span><span class="si">{</span><span class="n">cache_key</span><span class="si">}</span><span class="s2">&quot;</span><span class="p">,</span> <span class="mi">3600</span><span class="p">,</span> <span class="n">response</span><span class="p">)</span>
<span class="k">return</span> <span class="n">response</span>
</code></pre></div>
<hr />
<h3 id="4-observability">4. <strong>Моніторинг та observability</strong><a class="headerlink" href="#4-observability" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: Як зрозуміти, що відбувається з 1000 агентами?<br />
<strong>Рішення</strong>:</p>
<h4 id="a-prometheus-grafana">A. Метрики (Prometheus + Grafana)<a class="headerlink" href="#a-prometheus-grafana" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">prometheus_client</span><span class="w"> </span><span class="kn">import</span> <span class="n">Counter</span><span class="p">,</span> <span class="n">Histogram</span><span class="p">,</span> <span class="n">Gauge</span>
<span class="n">agent_requests</span> <span class="o">=</span> <span class="n">Counter</span><span class="p">(</span><span class="s1">&#39;greenfood_agent_requests&#39;</span><span class="p">,</span> <span class="s1">&#39;Запити до агентів&#39;</span><span class="p">,</span> <span class="p">[</span><span class="s1">&#39;agent_id&#39;</span><span class="p">])</span>
<span class="n">agent_latency</span> <span class="o">=</span> <span class="n">Histogram</span><span class="p">(</span><span class="s1">&#39;greenfood_agent_latency&#39;</span><span class="p">,</span> <span class="s1">&#39;Затримка агентів&#39;</span><span class="p">,</span> <span class="p">[</span><span class="s1">&#39;agent_id&#39;</span><span class="p">])</span>
<span class="n">active_agents</span> <span class="o">=</span> <span class="n">Gauge</span><span class="p">(</span><span class="s1">&#39;greenfood_active_agents&#39;</span><span class="p">,</span> <span class="s1">&#39;Активні агенти&#39;</span><span class="p">)</span>
<span class="c1"># Використання</span>
<span class="nd">@track_metrics</span>
<span class="k">def</span><span class="w"> </span><span class="nf">execute_agent</span><span class="p">(</span><span class="n">agent_id</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">task</span><span class="p">:</span> <span class="nb">dict</span><span class="p">):</span>
<span class="n">agent_requests</span><span class="o">.</span><span class="n">labels</span><span class="p">(</span><span class="n">agent_id</span><span class="o">=</span><span class="n">agent_id</span><span class="p">)</span><span class="o">.</span><span class="n">inc</span><span class="p">()</span>
<span class="k">with</span> <span class="n">agent_latency</span><span class="o">.</span><span class="n">labels</span><span class="p">(</span><span class="n">agent_id</span><span class="o">=</span><span class="n">agent_id</span><span class="p">)</span><span class="o">.</span><span class="n">time</span><span class="p">():</span>
<span class="k">return</span> <span class="n">agent</span><span class="o">.</span><span class="n">execute</span><span class="p">(</span><span class="n">task</span><span class="p">)</span>
</code></pre></div>
<h4 id="b-distributed-tracing-jaeger-tempo">B. Distributed Tracing (Jaeger / Tempo)<a class="headerlink" href="#b-distributed-tracing-jaeger-tempo" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">opentelemetry</span><span class="w"> </span><span class="kn">import</span> <span class="n">trace</span>
<span class="n">tracer</span> <span class="o">=</span> <span class="n">trace</span><span class="o">.</span><span class="n">get_tracer</span><span class="p">(</span><span class="vm">__name__</span><span class="p">)</span>
<span class="nd">@tracer</span><span class="o">.</span><span class="n">start_as_current_span</span><span class="p">(</span><span class="s2">&quot;fulfill_order&quot;</span><span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">fulfill_order</span><span class="p">(</span><span class="n">order_data</span><span class="p">:</span> <span class="nb">dict</span><span class="p">):</span>
<span class="k">with</span> <span class="n">tracer</span><span class="o">.</span><span class="n">start_as_current_span</span><span class="p">(</span><span class="s2">&quot;warehouse.reserve&quot;</span><span class="p">):</span>
<span class="n">warehouse_agent</span><span class="o">.</span><span class="n">reserve_items</span><span class="p">(</span><span class="n">order_data</span><span class="p">)</span>
<span class="k">with</span> <span class="n">tracer</span><span class="o">.</span><span class="n">start_as_current_span</span><span class="p">(</span><span class="s2">&quot;logistics.create_route&quot;</span><span class="p">):</span>
<span class="n">logistics_agent</span><span class="o">.</span><span class="n">create_route</span><span class="p">(</span><span class="n">order_data</span><span class="p">)</span>
</code></pre></div>
<h4 id="c-elk-stack-loki">C. Логування (ELK Stack / Loki)<a class="headerlink" href="#c-elk-stack-loki" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">import</span><span class="w"> </span><span class="nn">structlog</span>
<span class="n">logger</span> <span class="o">=</span> <span class="n">structlog</span><span class="o">.</span><span class="n">get_logger</span><span class="p">()</span>
<span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span>
<span class="s2">&quot;agent_execution_started&quot;</span><span class="p">,</span>
<span class="n">agent_id</span><span class="o">=</span><span class="s2">&quot;warehouse_agent&quot;</span><span class="p">,</span>
<span class="n">task_id</span><span class="o">=</span><span class="s2">&quot;task_123&quot;</span><span class="p">,</span>
<span class="n">vendor_id</span><span class="o">=</span><span class="s2">&quot;vendor_456&quot;</span><span class="p">,</span>
<span class="p">)</span>
</code></pre></div>
<hr />
<h3 id="5">5. <strong>Ізоляція та безпека</strong><a class="headerlink" href="#5" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: Один збійний агент не має зупинити всю систему<br />
<strong>Рішення</strong>:</p>
<h4 id="a-circuit-breaker-pattern">A. Circuit Breaker Pattern<a class="headerlink" href="#a-circuit-breaker-pattern" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">pybreaker</span><span class="w"> </span><span class="kn">import</span> <span class="n">CircuitBreaker</span>
<span class="n">breaker</span> <span class="o">=</span> <span class="n">CircuitBreaker</span><span class="p">(</span><span class="n">fail_max</span><span class="o">=</span><span class="mi">5</span><span class="p">,</span> <span class="n">timeout_duration</span><span class="o">=</span><span class="mi">60</span><span class="p">)</span>
<span class="nd">@breaker</span>
<span class="k">def</span><span class="w"> </span><span class="nf">call_warehouse_agent</span><span class="p">(</span><span class="n">task</span><span class="p">):</span>
<span class="c1"># Якщо warehouse_agent падає 5 разів підряд,</span>
<span class="c1"># circuit breaker відкривається на 60 сек</span>
<span class="k">return</span> <span class="n">warehouse_agent</span><span class="o">.</span><span class="n">execute</span><span class="p">(</span><span class="n">task</span><span class="p">)</span>
</code></pre></div>
<h4 id="b-sandbox">B. Sandbox для кожного агента<a class="headerlink" href="#b-sandbox" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="c1"># Docker container per agent instance</span>
<span class="n">docker</span> <span class="n">run</span> <span class="o">--</span><span class="n">rm</span> \
<span class="o">--</span><span class="n">memory</span><span class="o">=</span><span class="s2">&quot;512m&quot;</span> \
<span class="o">--</span><span class="n">cpus</span><span class="o">=</span><span class="s2">&quot;0.5&quot;</span> \
<span class="o">--</span><span class="n">name</span> <span class="n">agent_warehouse_001</span> \
<span class="n">greenfood</span><span class="o">/</span><span class="n">agent</span><span class="p">:</span><span class="n">latest</span> \
<span class="n">python</span> <span class="o">-</span><span class="n">m</span> <span class="n">greenfood</span><span class="o">.</span><span class="n">crew</span><span class="o">.</span><span class="n">warehouse_agent</span>
</code></pre></div>
<hr />
<h3 id="6-api">6. <strong>Дані та API</strong><a class="headerlink" href="#6-api" title="Permanent link">&para;</a></h3>
<p><strong>Проблема</strong>: 1000 агентів = тисячі запитів до БД/API<br />
<strong>Рішення</strong>:</p>
<h4 id="a-connection-pooling">A. Connection Pooling<a class="headerlink" href="#a-connection-pooling" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">sqlalchemy.pool</span><span class="w"> </span><span class="kn">import</span> <span class="n">QueuePool</span>
<span class="n">engine</span> <span class="o">=</span> <span class="n">create_engine</span><span class="p">(</span>
<span class="s2">&quot;postgresql://...&quot;</span><span class="p">,</span>
<span class="n">poolclass</span><span class="o">=</span><span class="n">QueuePool</span><span class="p">,</span>
<span class="n">pool_size</span><span class="o">=</span><span class="mi">20</span><span class="p">,</span>
<span class="n">max_overflow</span><span class="o">=</span><span class="mi">50</span><span class="p">,</span>
<span class="p">)</span>
</code></pre></div>
<h4 id="b-read-replicas">B. Read Replicas<a class="headerlink" href="#b-read-replicas" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="c1"># PostgreSQL з реплікацією</span>
<span class="nt">primary</span><span class="p">:</span>
<span class="w"> </span><span class="nt">host</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">postgres-primary</span>
<span class="w"> </span><span class="nt">port</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">5432</span>
<span class="nt">replicas</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">host</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">postgres-replica-1</span>
<span class="w"> </span><span class="nt">port</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">5432</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">host</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">postgres-replica-2</span>
<span class="w"> </span><span class="nt">port</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">5432</span>
<span class="c1"># Читання з реплік, запис в primary</span>
</code></pre></div>
<h4 id="c-api">C. Кешування на рівні API<a class="headerlink" href="#c-api" title="Permanent link">&para;</a></h4>
<div class="codehilite"><pre><span></span><code><span class="kn">from</span><span class="w"> </span><span class="nn">cachetools</span><span class="w"> </span><span class="kn">import</span> <span class="n">TTLCache</span>
<span class="n">api_cache</span> <span class="o">=</span> <span class="n">TTLCache</span><span class="p">(</span><span class="n">maxsize</span><span class="o">=</span><span class="mi">10000</span><span class="p">,</span> <span class="n">ttl</span><span class="o">=</span><span class="mi">300</span><span class="p">)</span>
<span class="nd">@cached</span><span class="p">(</span><span class="n">cache</span><span class="o">=</span><span class="n">api_cache</span><span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_product_catalog</span><span class="p">(</span><span class="n">vendor_id</span><span class="p">:</span> <span class="nb">str</span><span class="p">):</span>
<span class="k">return</span> <span class="n">db</span><span class="o">.</span><span class="n">query</span><span class="p">(</span><span class="n">Product</span><span class="p">)</span><span class="o">.</span><span class="n">filter_by</span><span class="p">(</span><span class="n">vendor_id</span><span class="o">=</span><span class="n">vendor_id</span><span class="p">)</span><span class="o">.</span><span class="n">all</span><span class="p">()</span>
</code></pre></div>
<hr />
<h2 id="_2">📊 Орієнтовні потужності<a class="headerlink" href="#_2" title="Permanent link">&para;</a></h2>
<h3 id="13">Поточна архітектура (13 агентів)<a class="headerlink" href="#13" title="Permanent link">&para;</a></h3>
<ul>
<li><strong>Throughput</strong>: ~10-20 задач/хв</li>
<li><strong>Latency</strong>: 3-10 сек на задачу</li>
<li><strong>Concurrent agents</strong>: 5-10</li>
</ul>
<h3 id="1000">Для 1000 агентів<a class="headerlink" href="#1000" title="Permanent link">&para;</a></h3>
<ul>
<li><strong>Throughput</strong>: 1000-5000 задач/хв</li>
<li><strong>Latency</strong>: &lt;5 сек (95 percentile)</li>
<li><strong>Concurrent agents</strong>: 500-1000</li>
</ul>
<h3 id="_3">Необхідна інфраструктура<a class="headerlink" href="#_3" title="Permanent link">&para;</a></h3>
<table>
<thead>
<tr>
<th>Компонент</th>
<th>Мінімум</th>
<th>Рекомендовано</th>
</tr>
</thead>
<tbody>
<tr>
<td><strong>LLM Nodes</strong></td>
<td>3x Ollama (RTX 4090)</td>
<td>10x Ollama або хмара</td>
</tr>
<tr>
<td><strong>Redis</strong></td>
<td>1 instance (16GB RAM)</td>
<td>Redis Cluster (64GB RAM)</td>
</tr>
<tr>
<td><strong>PostgreSQL</strong></td>
<td>1 primary + 1 replica</td>
<td>1 primary + 3 replicas</td>
</tr>
<tr>
<td><strong>Message Queue</strong></td>
<td>NATS Core</td>
<td>NATS JetStream Cluster</td>
</tr>
<tr>
<td><strong>API Gateway</strong></td>
<td>2 instances</td>
<td>5+ instances (auto-scaling)</td>
</tr>
<tr>
<td><strong>Monitoring</strong></td>
<td>Prometheus + Grafana</td>
<td>Full observability stack</td>
</tr>
</tbody>
</table>
<h3 id="_4">Орієнтовний бюджет<a class="headerlink" href="#_4" title="Permanent link">&para;</a></h3>
<ul>
<li><strong>On-premise</strong>: $15,000-30,000 (обладнання) + $500-1000/міс (утримання)</li>
<li><strong>Cloud</strong>: $2,000-5,000/міс (залежно від навантаження)</li>
</ul>
<hr />
<h2 id="_5">🛠️ План масштабування<a class="headerlink" href="#_5" title="Permanent link">&para;</a></h2>
<h3 id="phase-1-proof-of-concept-current">Phase 1: Proof of Concept (Current)<a class="headerlink" href="#phase-1-proof-of-concept-current" title="Permanent link">&para;</a></h3>
<ul>
<li>✅ 13 агентів</li>
<li>✅ 4 crews</li>
<li>✅ Базова інтеграція з Router</li>
<li><strong>Навантаження</strong>: 1-10 комітентів</li>
</ul>
<h3 id="phase-2-production-ready-1-3">Phase 2: Production Ready (1-3 місяці)<a class="headerlink" href="#phase-2-production-ready-1-3" title="Permanent link">&para;</a></h3>
<ul>
<li>[ ] Черги завдань (Celery/NATS)</li>
<li>[ ] Redis для стану</li>
<li>[ ] Моніторинг (Prometheus)</li>
<li>[ ] Circuit breakers</li>
<li><strong>Навантаження</strong>: 10-50 комітентів, 100-200 задач/год</li>
</ul>
<h3 id="phase-3-scale-out-3-6">Phase 3: Scale Out (3-6 місяців)<a class="headerlink" href="#phase-3-scale-out-3-6" title="Permanent link">&para;</a></h3>
<ul>
<li>[ ] LLM кластер (3-5 nodes)</li>
<li>[ ] PostgreSQL реплікація</li>
<li>[ ] API rate limiting</li>
<li>[ ] Distributed tracing</li>
<li><strong>Навантаження</strong>: 50-200 комітентів, 1000+ задач/год</li>
</ul>
<h3 id="phase-4-enterprise-scale-6-12">Phase 4: Enterprise Scale (6-12 місяців)<a class="headerlink" href="#phase-4-enterprise-scale-6-12" title="Permanent link">&para;</a></h3>
<ul>
<li>[ ] Auto-scaling (Kubernetes)</li>
<li>[ ] Multi-region deployment</li>
<li>[ ] Advanced caching (CDN, edge)</li>
<li>[ ] ML для оптимізації маршрутизації</li>
<li><strong>Навантаження</strong>: 200-1000 комітентів, 5000+ задач/год</li>
</ul>
<hr />
<h2 id="_6">💡 Швидкий старт для масштабування<a class="headerlink" href="#_6" title="Permanent link">&para;</a></h2>
<h3 id="1_1">1. Додай черги (швидко)<a class="headerlink" href="#1_1" title="Permanent link">&para;</a></h3>
<div class="codehilite"><pre><span></span><code><span class="c1"># Install Redis</span>
docker<span class="w"> </span>run<span class="w"> </span>-d<span class="w"> </span>-p<span class="w"> </span><span class="m">6379</span>:6379<span class="w"> </span>redis:alpine
<span class="c1"># Install Celery</span>
pip<span class="w"> </span>install<span class="w"> </span>celery<span class="w"> </span>redis
</code></pre></div>
<div class="codehilite"><pre><span></span><code><span class="c1"># services/greenfood/tasks.py</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">celery</span><span class="w"> </span><span class="kn">import</span> <span class="n">Celery</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">services.greenfood.crew.greenfood_crews</span><span class="w"> </span><span class="kn">import</span> <span class="n">GREENFOOD_CREWS</span>
<span class="n">app</span> <span class="o">=</span> <span class="n">Celery</span><span class="p">(</span><span class="s1">&#39;greenfood&#39;</span><span class="p">,</span> <span class="n">broker</span><span class="o">=</span><span class="s1">&#39;redis://localhost:6379&#39;</span><span class="p">)</span>
<span class="nd">@app</span><span class="o">.</span><span class="n">task</span>
<span class="k">def</span><span class="w"> </span><span class="nf">execute_crew_async</span><span class="p">(</span><span class="n">crew_name</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">data</span><span class="p">:</span> <span class="nb">dict</span><span class="p">):</span>
<span class="n">crew</span> <span class="o">=</span> <span class="n">GREENFOOD_CREWS</span><span class="p">[</span><span class="n">crew_name</span><span class="p">]</span>
<span class="c1"># ... setup tasks ...</span>
<span class="k">return</span> <span class="n">crew</span><span class="o">.</span><span class="n">kickoff</span><span class="p">()</span>
</code></pre></div>
<h3 id="2">2. Додай моніторинг (швидко)<a class="headerlink" href="#2" title="Permanent link">&para;</a></h3>
<div class="codehilite"><pre><span></span><code><span class="c1"># Prometheus + Grafana</span>
docker-compose<span class="w"> </span>up<span class="w"> </span>-d<span class="w"> </span>prometheus<span class="w"> </span>grafana
</code></pre></div>
<h3 id="3-load-balancer-llm">3. Додай load balancer для LLM (середньо)<a class="headerlink" href="#3-load-balancer-llm" title="Permanent link">&para;</a></h3>
<div class="codehilite"><pre><span></span><code><span class="c1"># nginx.conf</span>
<span class="l l-Scalar l-Scalar-Plain">upstream ollama_backend {</span>
<span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">least_conn;</span><span class="w"> </span><span class="c1"># Балансування по найменшому навантаженню</span>
<span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">server ollama-1:11434;</span>
<span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">server ollama-2:11434;</span>
<span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">server ollama-3:11434;</span>
<span class="l l-Scalar l-Scalar-Plain">}</span>
</code></pre></div>
<hr />
<h2 id="_7">🎯 Висновок<a class="headerlink" href="#_7" title="Permanent link">&para;</a></h2>
<p><strong>Для 1000 агентів потрібно</strong>:
1. ✅ Архітектура вже хороша (crewAI + модульність)
2. ⏳ Додати черги та state management (1-2 тижні)
3. ⏳ Масштабувати LLM інфраструктуру (2-4 тижні)
4. ⏳ Додати observability (1-2 тижні)
5. ⏳ Тестування під навантаженням (2-4 тижні)</p>
<p><strong>Реальний термін до production-ready 1000 агентів: 2-3 місяці</strong></p>
<hr />
<p><em>Документ створено: 2025-11-18</em></p>
</article>
</div>
<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
</div>
</main>
<footer class="md-footer">
<div class="md-footer-meta md-typeset">
<div class="md-footer-meta__inner md-grid">
<div class="md-copyright">
Made with
<a href="https://squidfunk.github.io/mkdocs-material/" target="_blank" rel="noopener">
Material for MkDocs
</a>
</div>
</div>
</div>
</footer>
</div>
<div class="md-dialog" data-md-component="dialog">
<div class="md-dialog__inner md-typeset"></div>
</div>
<script id="__config" type="application/json">{"base": "../..", "features": ["navigation.sections", "navigation.instant", "content.code.copy"], "search": "../../assets/javascripts/workers/search.b8dbb3d2.min.js", "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}}</script>
<script src="../../assets/javascripts/bundle.3220b9d7.min.js"></script>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink