SlideShare a Scribd company logo

Serverless Clouds (FaaS) and request context isolation in Node.js

Timur Shemsedinov, profile picture
Timur Shemsedinov

The document discusses serverless computing and techniques for isolating request contexts in Node.js applications. It covers infrastructure types like bare metal, virtualization, containers and serverless platforms. It also discusses different levels of isolation for Node.js including hardware segmentation, virtual machines, containers, processes, threads and software contexts. The document advocates for a service-oriented architecture with microservices and techniques like software isolation to achieve request context isolation for Node.js applications.

Software
Related topics:
Node.js Development
1 of 32
Downloaded 11 times
1
2
3
4
5
6
7
8
9
10
Most read
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Most read
26
27
28
29
30
31
32
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c toString().padStart(width) : cell.padEnd(width); }).join(''))).join }; const proportion = (max, val) => Math.round(val * 100 / max); co calcProportion = table => { table.sort((row1, row2) => row2[DENSITY row1[DENSITY_COL]); const maxDensity = table[0][DENSITY_COL]; table forEach(row => { row.push(proportion(maxDensity, row[DENSITY_COL])) return table; }; const getDataset = file => { const lines = fs.read FileSync(file, 'utf8').toString().split('n'); lines.shift(); lines return lines.map(line => line.split(',')); }; const main = compose (getDataset, calcProportion, renderTable); const fs = require('fs' compose = (...funcs) => x => funcs.reduce((x, fn) => fn(x), x); con DENSITY_COL = 3; const renderTable = table => { const cellWidth = [ 8, 8, 18, 6]; return table.map(row => (row.map((cell, i) => { const = cellWidth[i]; return i ? cell.toString().padStart(width) : cell.p (width); }).join(''))).join('n'); }; const proportion = (max, val) Serverless Clouds и изоляция контекстов запросов в Node.js Тимур Шемсединов github.com/HowProgrammingWorks github.com/tshemsedinov Chief Software Architect at Metarhia Lecturer at Kiev Polytechnic Institute
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Clouds vs App Clouds Hardware Infrastructure Application State Functions Hardware Infrastructure Application X Functions IaaSCloud FaaSCloud
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Types Infra Runtime Storage ● Bare metal none stateful stateful ● Shared hosting slice stateless stateful ● Virtualization slice stateful stateful ● Containerization slice stateful stateful ● Clusterization join stateful stateless ● Serverless join stateless stateless
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Aspects Sec Lock TTM ● Bare metal ++++ no + ● Shared hosting + tech ++ ● Virtualization +++ no +++ ● Containerization +++ tech +++ ● Clusterization +++ tech ++++ ● Serverless ++ vendor +++++
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Benefits ● Service price? (evangelists told us...) ● Efficiency: Performance? Speed? Latency? ● Easy to test, deploy, maintain? ● Security? Reliability? Flexibility? Quality? ● Quick development? ● Reduces development cost? ● Scalability?
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What we pay for? We pay for: ● lack of available professionals ● lack of competencies ● lack of available technologies ● lack of funding for our projects ● lack of time
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What can be automated? Yes ● Infrastructure ● Monitoring ● Networking ● Integration (CI) ● Deployment (CD) No ● Scalability, sharding ● Performance ● Quality ● Security ● Interactivity
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Cost Optimization Cases ● Small services, sometimes cold Can reduce cost x10 (great: $10 to $1) ● Highload >100k online, always warm Single bare metal can hold load Try to calculate serverless cost...
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Disadvantages ● High resource consumption ● Stateless nature and no application integrity ● Interactivity issue (separate solution needed) ● Development and debug issues ● Deploy and maintain issues ● Vendor lock, not open source ● Where is no promised simplicity
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Middleware Madness router.get('/user/:id', (req, res, next) => { const id = parseInt(req.params.id); const query = 'SELECT * FROM users WHERE id = $1'; pool.query(query, [id], (err, data) => { if (err) throw err; res.status(200).json(data.rows); next(); }); });
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Code Structure and Patterns exports.handler = (event, context, callback) => { const { Client } = require('pg'); const client = new Client(); client.connect(); const id = parseInt(event.pathParameters.id); const query = 'SELECT * FROM users WHERE id = $1'; client.query(query, [id], (err, data) => { callback(err, { statusCode: 200, body: JSON.stringify(data.rows)}); }); };
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? async (arg1, arg2, arg3) => { const [data1, data2] = await Promise.all( [getData(arg1), getData(arg2)] ); const data3 = await getData(arg3); if (!data3) throw new Error('Message'); return await processData(data1, data2, data3); }
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? async (arg1, arg2, arg3) => { const data1 = await getData(arg1); if (!data1) throw new Error('Message'); const [data2, data3] = await Promise.all( [getData(arg2), getData(arg3)] ); return await processData(data1, data2, data3); }
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Equivalent example async id => await application.database .select('users') .where({ id });
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Complex Query async id => await application.database .select('users') .where({ id }) .cache({ timeout: 30000, invalidate: { id }) .projection({ name: ['name', toUpperCase], age: ['birth', toAge], place: ['address', getCity, getGeocode], });
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Layered Architecture Server-side ● Layered ● Microservices ● Serverless Database Data Access Layer Business-logic API Network Client UI
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? ● Apps consolidation ● Stateful cloud applications ● Interactivity (Websockets, TCP, TLS support) ● No vendor lock ● Private clouds ● Do not overpay for clouds
● Architecture and layered approach ● Async I/O for business-logic parallelization ● Long-lived processes: in-memory, reuse ● Server inside application (not vice versa) ● Minimize IPC and serialization ● Open source ● But we need request isolation const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c How do we achieve this?
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Why Isolation? ● Errors, crashes ● Memory leaks and other resources ● Application: data, database connections ● File system and root directory ● OS environment, PID, IPC ● OS Security: users, groups ● Networking: socket descriptors, ports, hosts
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation: Virtualization Hardware OS Hypervisor OS App OS App OS App OS App
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation: Containerization Hardware OS Containers App App App App App App App
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Microisolation Hardware OS App Server App App App App App App App App App AppAppAppAppApp AppApp
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation Levels (for Node.js) ● Hardware segmentation, Servers ● Virtual machine (hypervisor) ● Container (docker) ● Process (node) ● Thread (worker_threads) ● Sandbox (vm.createContext, vm.Script) ● Software context (object, closure)
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c SOA Architecture DB API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network SingleServer SOAArchitecture
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Microservices Architecture DB API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network DB DB DB SingleServer Microservices
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network SingleServer Serverless
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Metaserverless Experiments ● Application is not a separate functions, application has distributed in-memory state ● Functions can be executed sequentially and parallely in asynchronous style ● Applications have long life and structure ● Interactivity (Websockets, TCP, TLS support) ● No vendor lock, Private clouds, Open Source
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Vendor Lock Prevention Checklist ● Wrap vendor services ● Concentrate on architecture: layers ● Code quality and competencies ● Think twice before following hype and trends ● Remove dependencies if possible
const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Vendor Loyalty Checklisk ● Use everything as a service ● Follow guidelines ● Сut risky developments ● Relax ● Share your income

More Related Content

PDF
InnoDB MVCC Architecture (by 권건우)
I Goo Lee.
 
PDF
Varnashrama dharma
Mamatha254162
 
PDF
KINH ĐỊA TẠNG BỒ-TÁT BỔN NGUYỆN
Chiến Thắng Bản Thân
 
PDF
PostgreSQL, performance for queries with grouping
Alexey Bashtanov
 
PDF
ClickHouse Query Performance Tips and Tricks, by Robert Hodges, Altinity CEO
Altinity Ltd
 
PDF
Базы данных в 2020
Timur Shemsedinov
 
PDF
Planejador de Consultas do PostgreSQL
Fabrízio Mello
 
PPTX
Wk 7 chinese buddhism
Roxanne Farrar
 
InnoDB MVCC Architecture (by 권건우)
I Goo Lee.
 
Varnashrama dharma
Mamatha254162
 
KINH ĐỊA TẠNG BỒ-TÁT BỔN NGUYỆN
Chiến Thắng Bản Thân
 
PostgreSQL, performance for queries with grouping
Alexey Bashtanov
 
ClickHouse Query Performance Tips and Tricks, by Robert Hodges, Altinity CEO
Altinity Ltd
 
Базы данных в 2020
Timur Shemsedinov
 
Planejador de Consultas do PostgreSQL
Fabrízio Mello
 
Wk 7 chinese buddhism
Roxanne Farrar
 

Similar to Serverless Clouds (FaaS) and request context isolation in Node.js (20)

PDF
Private cloud without vendor lock // Serverless
Timur Shemsedinov
 
PDF
How to keep control and safety in the clouds
Timur Shemsedinov
 
PDF
Web Locks API
Timur Shemsedinov
 
PDF
Race-conditions-web-locks-and-shared-memory
Timur Shemsedinov
 
PDF
How are Race Conditions in single threaded JavaScript possible?
Timur Shemsedinov
 
PDF
Node.js in 2020
Timur Shemsedinov
 
PDF
Node.js in 2020 - part 1
Timur Shemsedinov
 
PDF
Asynchronous programming and mutlithreading
Timur Shemsedinov
 
PDF
Node.js middleware: Never again!
Timur Shemsedinov
 
PDF
Новое в JavaScript: ES.Next, ECMAScript 2020, ES11, ES10, ES9, ES8, ES7, ES6,...
Timur Shemsedinov
 
PDF
Node.js in 2020 - part 3
Timur Shemsedinov
 
PDF
JavaScript в браузере: Web API (часть 1)
Timur Shemsedinov
 
PDF
Node.js in 2020 - part 2
Timur Shemsedinov
 
PDF
Metarhia KievJS 22-Feb-2018
Timur Shemsedinov
 
PDF
NoSQL meets Microservices - Michael Hackstein
distributed matters
 
PDF
Using Graphs for Feature Engineering_ Graph Reduce-2.pdf
Wes Madrigal
 
PDF
OrientDB: Unlock the Value of Document Data Relationships
Fabrizio Fortino
 
PDF
MongoDB in FS
MongoDB
 
PDF
Building a Microservices-based ERP System
MongoDB
 
PDF
JS Fest 2019 Node.js Antipatterns
Timur Shemsedinov
 
Private cloud without vendor lock // Serverless
Timur Shemsedinov
 
How to keep control and safety in the clouds
Timur Shemsedinov
 
Web Locks API
Timur Shemsedinov
 
Race-conditions-web-locks-and-shared-memory
Timur Shemsedinov
 
How are Race Conditions in single threaded JavaScript possible?
Timur Shemsedinov
 
Node.js in 2020
Timur Shemsedinov
 
Node.js in 2020 - part 1
Timur Shemsedinov
 
Asynchronous programming and mutlithreading
Timur Shemsedinov
 
Node.js middleware: Never again!
Timur Shemsedinov
 
Новое в JavaScript: ES.Next, ECMAScript 2020, ES11, ES10, ES9, ES8, ES7, ES6,...
Timur Shemsedinov
 
Node.js in 2020 - part 3
Timur Shemsedinov
 
JavaScript в браузере: Web API (часть 1)
Timur Shemsedinov
 
Node.js in 2020 - part 2
Timur Shemsedinov
 
Metarhia KievJS 22-Feb-2018
Timur Shemsedinov
 
NoSQL meets Microservices - Michael Hackstein
distributed matters
 
Using Graphs for Feature Engineering_ Graph Reduce-2.pdf
Wes Madrigal
 
OrientDB: Unlock the Value of Document Data Relationships
Fabrizio Fortino
 
MongoDB in FS
MongoDB
 
Building a Microservices-based ERP System
MongoDB
 
JS Fest 2019 Node.js Antipatterns
Timur Shemsedinov
 
Ad

More from Timur Shemsedinov (17)

PDF
How to use Chat GPT in JavaScript optimizations for Node.js
Timur Shemsedinov
 
PDF
IT Revolution in 2023-2024: AI, GPT, business transformation, future professi...
Timur Shemsedinov
 
PDF
Multithreading in Node.js and JavaScript
Timur Shemsedinov
 
PDF
Node.js threads for I/O-bound tasks
Timur Shemsedinov
 
PDF
Node.js Меньше сложности, больше надежности Holy.js 2021
Timur Shemsedinov
 
PDF
Rethinking low-code
Timur Shemsedinov
 
PDF
Hat full of developers
Timur Shemsedinov
 
PDF
FwDays 2021: Metarhia Technology Stack for Node.js
Timur Shemsedinov
 
PDF
Node.js for enterprise - JS Conference
Timur Shemsedinov
 
PDF
Node.js for enterprise 2021 - JavaScript Fwdays 3
Timur Shemsedinov
 
PDF
Node.js in 2021
Timur Shemsedinov
 
PDF
Patterns and antipatterns
Timur Shemsedinov
 
PDF
Information system structure and architecture
Timur Shemsedinov
 
PDF
Введение в SQL
Timur Shemsedinov
 
PDF
Почему хорошее ИТ-образование невостребовано рыночком
Timur Shemsedinov
 
PDF
Node.js security
Timur Shemsedinov
 
PDF
Prototype programming in JavaScript
Timur Shemsedinov
 
How to use Chat GPT in JavaScript optimizations for Node.js
Timur Shemsedinov
 
IT Revolution in 2023-2024: AI, GPT, business transformation, future professi...
Timur Shemsedinov
 
Multithreading in Node.js and JavaScript
Timur Shemsedinov
 
Node.js threads for I/O-bound tasks
Timur Shemsedinov
 
Node.js Меньше сложности, больше надежности Holy.js 2021
Timur Shemsedinov
 
Rethinking low-code
Timur Shemsedinov
 
Hat full of developers
Timur Shemsedinov
 
FwDays 2021: Metarhia Technology Stack for Node.js
Timur Shemsedinov
 
Node.js for enterprise - JS Conference
Timur Shemsedinov
 
Node.js for enterprise 2021 - JavaScript Fwdays 3
Timur Shemsedinov
 
Node.js in 2021
Timur Shemsedinov
 
Patterns and antipatterns
Timur Shemsedinov
 
Information system structure and architecture
Timur Shemsedinov
 
Введение в SQL
Timur Shemsedinov
 
Почему хорошее ИТ-образование невостребовано рыночком
Timur Shemsedinov
 
Node.js security
Timur Shemsedinov
 
Prototype programming in JavaScript
Timur Shemsedinov
 
Ad

Recently uploaded (20)

PDF
Flood Susceptibility Mapping Using Image-Based 2D-CNN Deep Learnin. Overview ...
lawrenceomai2
 
PPTX
Online Work Permit System for Fast Permit Processing
SHEQ Network Limited
 
PPTX
Odoo POS Development Services by CandidRoot Solutions
CandidRoot Solutions Private Limited
 
PPTX
CHAPTER 12 - CYBER SECURITY AND FUTURE SKILLS (1) (1).pptx
AnujKumar530915
 
PPTX
Agentic AI Use Case- Contract Lifecycle Management (CLM).pptx
mhxyzzp
 
PPTX
Transform Your Business with a Software ERP System
Canada Software Company
 
PDF
How Creative Agencies Leverage Project Management Software.pdf
Orangescrum
 
PDF
Odoo Companies in India – Driving Business Transformation.pdf
azhuhardeen1968
 
PDF
top salesforce developer skills in 2025.pdf
CloudMetic
 
PPTX
Oracle E-Business Suite: A Comprehensive Guide for Modern Enterprises
Conacent Consulting
 
PDF
Wondershare Filmora 15 Crack With Activation Key [2025
ghrom2211g
 
PDF
How to Choose the Right IT Partner for Your Business in Malaysia
Asian Business Services & Technologies
 
PPTX
Lecture 3: Operating Systems Introduction to Computer Hardware Systems
hci7se
 
PDF
Internet Downloader Manager (IDM) Crack 6.42 Build 42 Updates Latest 2025
itskinga12
 
PDF
Claude Code: Everyone is a 10x Developer - A Comprehensive AI-Powered CLI Tool
William Chong
 
PDF
Which alternative to Crystal Reports is best for small or large businesses.pdf
Varsha Nayak
 
PDF
Design an Analysis of Algorithms II-SECS-1021-03
DilanEscobar1
 
PDF
Audit Checklist Design Aligning with ISO, IATF, and Industry Standards — Omne...
rohnjim07
 
PDF
Why TechBuilder is the Future of Pickup and Delivery App Development (1).pdf
TechBuilder
 
PPTX
VVF-Customer-Presentation2025-Ver1.9.pptx
blackmambaettijean
 
Flood Susceptibility Mapping Using Image-Based 2D-CNN Deep Learnin. Overview ...
lawrenceomai2
 
Online Work Permit System for Fast Permit Processing
SHEQ Network Limited
 
Odoo POS Development Services by CandidRoot Solutions
CandidRoot Solutions Private Limited
 
CHAPTER 12 - CYBER SECURITY AND FUTURE SKILLS (1) (1).pptx
AnujKumar530915
 
Agentic AI Use Case- Contract Lifecycle Management (CLM).pptx
mhxyzzp
 
Transform Your Business with a Software ERP System
Canada Software Company
 
How Creative Agencies Leverage Project Management Software.pdf
Orangescrum
 
Odoo Companies in India – Driving Business Transformation.pdf
azhuhardeen1968
 
top salesforce developer skills in 2025.pdf
CloudMetic
 
Oracle E-Business Suite: A Comprehensive Guide for Modern Enterprises
Conacent Consulting
 
Wondershare Filmora 15 Crack With Activation Key [2025
ghrom2211g
 
How to Choose the Right IT Partner for Your Business in Malaysia
Asian Business Services & Technologies
 
Lecture 3: Operating Systems Introduction to Computer Hardware Systems
hci7se
 
Internet Downloader Manager (IDM) Crack 6.42 Build 42 Updates Latest 2025
itskinga12
 
Claude Code: Everyone is a 10x Developer - A Comprehensive AI-Powered CLI Tool
William Chong
 
Which alternative to Crystal Reports is best for small or large businesses.pdf
Varsha Nayak
 
Design an Analysis of Algorithms II-SECS-1021-03
DilanEscobar1
 
Audit Checklist Design Aligning with ISO, IATF, and Industry Standards — Omne...
rohnjim07
 
Why TechBuilder is the Future of Pickup and Delivery App Development (1).pdf
TechBuilder
 
VVF-Customer-Presentation2025-Ver1.9.pptx
blackmambaettijean
 

Serverless Clouds (FaaS) and request context isolation in Node.js

  • 1. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c toString().padStart(width) : cell.padEnd(width); }).join(''))).join }; const proportion = (max, val) => Math.round(val * 100 / max); co calcProportion = table => { table.sort((row1, row2) => row2[DENSITY row1[DENSITY_COL]); const maxDensity = table[0][DENSITY_COL]; table forEach(row => { row.push(proportion(maxDensity, row[DENSITY_COL])) return table; }; const getDataset = file => { const lines = fs.read FileSync(file, 'utf8').toString().split('n'); lines.shift(); lines return lines.map(line => line.split(',')); }; const main = compose (getDataset, calcProportion, renderTable); const fs = require('fs' compose = (...funcs) => x => funcs.reduce((x, fn) => fn(x), x); con DENSITY_COL = 3; const renderTable = table => { const cellWidth = [ 8, 8, 18, 6]; return table.map(row => (row.map((cell, i) => { const = cellWidth[i]; return i ? cell.toString().padStart(width) : cell.p (width); }).join(''))).join('n'); }; const proportion = (max, val) Serverless Clouds и изоляция контекстов запросов в Node.js Тимур Шемсединов github.com/HowProgrammingWorks github.com/tshemsedinov Chief Software Architect at Metarhia Lecturer at Kiev Polytechnic Institute
  • 2. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Clouds vs App Clouds Hardware Infrastructure Application State Functions Hardware Infrastructure Application X Functions IaaSCloud FaaSCloud
  • 3. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Types Infra Runtime Storage ● Bare metal none stateful stateful ● Shared hosting slice stateless stateful ● Virtualization slice stateful stateful ● Containerization slice stateful stateful ● Clusterization join stateful stateless ● Serverless join stateless stateless
  • 4. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Infrastructure Aspects Sec Lock TTM ● Bare metal ++++ no + ● Shared hosting + tech ++ ● Virtualization +++ no +++ ● Containerization +++ tech +++ ● Clusterization +++ tech ++++ ● Serverless ++ vendor +++++
  • 5. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Benefits ● Service price? (evangelists told us...) ● Efficiency: Performance? Speed? Latency? ● Easy to test, deploy, maintain? ● Security? Reliability? Flexibility? Quality? ● Quick development? ● Reduces development cost? ● Scalability?
  • 6. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What we pay for? We pay for: ● lack of available professionals ● lack of competencies ● lack of available technologies ● lack of funding for our projects ● lack of time
  • 7. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What can be automated? Yes ● Infrastructure ● Monitoring ● Networking ● Integration (CI) ● Deployment (CD) No ● Scalability, sharding ● Performance ● Quality ● Security ● Interactivity
  • 8. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Cost Optimization Cases ● Small services, sometimes cold Can reduce cost x10 (great: $10 to $1) ● Highload >100k online, always warm Single bare metal can hold load Try to calculate serverless cost...
  • 9. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Disadvantages ● High resource consumption ● Stateless nature and no application integrity ● Interactivity issue (separate solution needed) ● Development and debug issues ● Deploy and maintain issues ● Vendor lock, not open source ● Where is no promised simplicity
  • 10. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Middleware Madness router.get('/user/:id', (req, res, next) => { const id = parseInt(req.params.id); const query = 'SELECT * FROM users WHERE id = $1'; pool.query(query, [id], (err, data) => { if (err) throw err; res.status(200).json(data.rows); next(); }); });
  • 11. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Code Structure and Patterns exports.handler = (event, context, callback) => { const { Client } = require('pg'); const client = new Client(); client.connect(); const id = parseInt(event.pathParameters.id); const query = 'SELECT * FROM users WHERE id = $1'; client.query(query, [id], (err, data) => { callback(err, { statusCode: 200, body: JSON.stringify(data.rows)}); }); };
  • 12. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? async (arg1, arg2, arg3) => { const [data1, data2] = await Promise.all( [getData(arg1), getData(arg2)] ); const data3 = await getData(arg3); if (!data3) throw new Error('Message'); return await processData(data1, data2, data3); }
  • 13. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? async (arg1, arg2, arg3) => { const data1 = await getData(arg1); if (!data1) throw new Error('Message'); const [data2, data3] = await Promise.all( [getData(arg2), getData(arg3)] ); return await processData(data1, data2, data3); }
  • 14. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Equivalent example async id => await application.database .select('users') .where({ id });
  • 15. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Complex Query async id => await application.database .select('users') .where({ id }) .cache({ timeout: 30000, invalidate: { id }) .projection({ name: ['name', toUpperCase], age: ['birth', toAge], place: ['address', getCity, getGeocode], });
  • 16. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Layered Architecture Server-side ● Layered ● Microservices ● Serverless Database Data Access Layer Business-logic API Network Client UI
  • 17. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c What do we want? ● Apps consolidation ● Stateful cloud applications ● Interactivity (Websockets, TCP, TLS support) ● No vendor lock ● Private clouds ● Do not overpay for clouds
  • 18. ● Architecture and layered approach ● Async I/O for business-logic parallelization ● Long-lived processes: in-memory, reuse ● Server inside application (not vice versa) ● Minimize IPC and serialization ● Open source ● But we need request isolation const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c How do we achieve this?
  • 19. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Why Isolation? ● Errors, crashes ● Memory leaks and other resources ● Application: data, database connections ● File system and root directory ● OS environment, PID, IPC ● OS Security: users, groups ● Networking: socket descriptors, ports, hosts
  • 20. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation: Virtualization Hardware OS Hypervisor OS App OS App OS App OS App
  • 21. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation: Containerization Hardware OS Containers App App App App App App App
  • 22. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Microisolation Hardware OS App Server App App App App App App App App App AppAppAppAppApp AppApp
  • 23. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Isolation Levels (for Node.js) ● Hardware segmentation, Servers ● Virtual machine (hypervisor) ● Container (docker) ● Process (node) ● Thread (worker_threads) ● Sandbox (vm.createContext, vm.Script) ● Software context (object, closure)
  • 24. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c SOA Architecture DB API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network SingleServer SOAArchitecture
  • 25. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Microservices Architecture DB API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network DB DB DB SingleServer Microservices
  • 26. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API + BL Network API + BL API + BL API + BL Database Data Access Layer Business-logic API Network SingleServer Serverless
  • 27. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
  • 28. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
  • 29. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Serverless Architecture Database API+BL Network Database Data Access Layer Business-logic API Network API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL API+BL SingleServer Serverless
  • 30. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Metaserverless Experiments ● Application is not a separate functions, application has distributed in-memory state ● Functions can be executed sequentially and parallely in asynchronous style ● Applications have long life and structure ● Interactivity (Websockets, TCP, TLS support) ● No vendor lock, Private clouds, Open Source
  • 31. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Vendor Lock Prevention Checklist ● Wrap vendor services ● Concentrate on architecture: layers ● Code quality and competencies ● Think twice before following hype and trends ● Remove dependencies if possible
  • 32. const fs = require('fs'); const compose = (...funcs) => x => funcs. reduce((x, fn) => fn(x), x); const DENSITY_COL = 3; const renderTab table => { const cellWidth = [18, 10, 8, 8, 18, 6]; return table.ma => (row.map((cell, i) => { const width = cellWidth[i]; return i ? c Vendor Loyalty Checklisk ● Use everything as a service ● Follow guidelines ● Сut risky developments ● Relax ● Share your income