System Design

Scalability asoslari

Scalability — tizimning yuklama ortishi bilan bajarilishini saqlash yoki yaxshilash qobiliyati.

Nega Scalability muhim?

Tasavvur qiling, siz startup yaratdingiz:

• 1-kun: 100 foydalanuvchi → bitta server yetadi
• 1-oy: 10,000 foydalanuvchi → server sekinlasha boshladi
• 6-oy: 1,000,000 foydalanuvchi → server ishlamay qoldi

Muvaffaqiyatli mahsulot o’sadi. O’sish uchun tizim scalable bo’lishi kerak.

Vertical Scaling (Vertikal kengayish)

Scale Up — mavjud serverni kuchliroq mashina bilan almashtirish.

Server 1.0 → Server 2.0
- 4 CPU   → 16 CPU
- 8 GB RAM → 64 GB RAM
- 500 GB disk → 2 TB SSD

Afzalliklari

• Oddiy implementatsiya: faqat serverni almashtirish kerak
• Murakkablik yo’q: arxitektura o’zgarmaydi
• Ma’lumotlar bir joyda: distributed muammolar yo’q

Kamchiliklari

• Cheklov bor: bitta mashinani cheksiz kuchaytira olmaysiz
• Single point of failure: server ishlamasa, hammasi to’xtaydi
• Qimmat: katta serverlar juda qimmat
• Downtime kerak: serverni almashtirganda to’xtatish kerak

Qachon ishlatiladi?

• Startup boshlang’ich bosqichida
• Database kabi stateful komponentlar uchun
• Legacy tizimlar uchun (qayta yozish qiyin bo’lsa)

Horizontal Scaling (Gorizontal kengayish)

Scale Out — ko’proq serverlar qo’shish.

Server 1 → Server 1, 2, 3, 4, 5, ...
- 4 CPU    - 4 CPU (har birida)
- 8 GB RAM - 8 GB RAM (har birida)

Afzalliklari

• Cheksiz kengayish: minglab server qo’sha olasiz
• High availability: bitta server ishlamasa, qolganlari ishlaydi
• Cost-effective: oddiy serverlar arzon
• Zero downtime: yangi server qo’shish paytida tizim ishlaydi

Kamchiliklari

• Murakkab: load balancer, distributed system kerak
• Ma’lumotlar taqsimlangan: consistency muammolari
• Dasturni o’zgartirish kerak: stateless qilish kerak
• Network overhead: serverlar o’rtasida aloqa kerak

Qachon ishlatiladi?

• Million+ foydalanuvchilar
• Microservices arxitektura
• Cloud-native ilovalar
• High availability talab qilinadigan tizimlar

Vertical vs Horizontal: Taqqoslash

Real-world misollar

Instagram

• Dastlab: Vertical scaling (katta EC2 instancelar)
• Hozir: Horizontal scaling (minglab server)
• Sabab: billion+ foydalanuvchilarga xizmat

Stack Overflow

• Hozir ham: Asosan vertical scaling
• Sabab: O’rta hajmdagi yuklama, sodda arxitektura yetarli

Netflix

• To’liq horizontal: 100,000+ microservice instancelari
• Sabab: global scale, high availability

Aralash yondashuv (Hybrid)

Ko’pincha ikkalasi birgalikda ishlatiladi:

┌─────────────────────┐
│  Load Balancer      │ ← Horizontal scaling
└─────────┬───────────┘
          │
     ┌────┴────┬────────────┐
     │         │            │
┌────▼────┐ ┌─▼────────┐ ┌─▼────────┐
│ Server 1│ │ Server 2 │ │ Server 3 │ ← Horizontal
│ (16 CPU)│ │ (16 CPU) │ │ (16 CPU) │ ← Vertical
└─────────┘ └──────────┘ └──────────┘

Har bir server kuchli (vertical) + Ko’p serverlar (horizontal)

Stateful vs Stateless

Horizontal scaling uchun stateless bo’lish muhim:

Stateful (Muammo)

User → Server 1 (session saqlanadi)
User → Server 2 (session yo'q!) 

Stateless (Yechim)

User → Server 1 → Redis (session tashqarida)
User → Server 2 → Redis (bir xil session) 

Auto-scaling

Modern cloud platformalar avtomatik scale qiladi:

// AWS Auto Scaling Rule
if (CPU > 70%) {
  add_server();
}
if (CPU < 30%) {
  remove_server();
}

Afzalliklari:

• Traffic spike’larga avtomatik javob
• Kam trafik vaqtida pul tejash
• Manual monitoring kerak emas

Xulosa

Scalability — muvaffaqiyatli tizimning asosi.

Qoidalar:

1. Kichik boshlaganda — vertical yetarli
2. O’sish kutilsa — horizontal rejalashtiring
3. Stateless arxitektura yarating
4. Auto-scaling o’rnating
5. Monitoring qo’ying (metrikalar kuzating)

Keyingi dars: Tarmoq asoslari va internet protokollari.