Apa kuwi distributed monolith lan anti-pattern microservices liyane?

Question

Accepted Answer

Asil terburuk saka microservices yaiku **"distributed monolith"**: layanan sing pisah sacara fisik nanging begitu erat dipasungake nganti kudu dideploy dhewe-dhewe — sampeyan entuk kabeh biaya distribusi lan ora ana independensi.

## Tandha-tandha distributed monolith

```text
✗ Changing one service forces redeploying several others
✗ Services share a database / shared schema
✗ A single request fans out through many synchronous hops
✗ Releases must be coordinated across teams
→ you pay network + ops cost but lost independent deployability
```

## Anti-pattern umum liyane

| Anti-pattern | Masalah |
|---|---|
| Shared database | Nggabung maneh layanan; ora bisa ngowahi skema kanthi independen |
| Chatty services | Akeh panggilan sync per panjaluk → latensi, kalemahan |
| Nano-services | Paling halus; overhead ngendang logika |
| No API versioning | Siji perubahan rusak saben konsumen |
| Synchronous chains | Latensi lan kegagalan kalilipan ing A→B→C→D |
| God service | Siji layanan sing kabeh gumantung (monolith anyar) |

## Panyebab utama

Biasane **wates sing lepat**. Layanan sing dipérang dening lapisan teknis utawa entitas (ora kapabilitas bisnis) bakal dadi chatty lan saling gumantung.

```text
bad:  User-svc, Product-svc, Cart-svc all called for one checkout
good: Checkout owns the capability end to end
```

## Mengapa iku penting

Distributed monolith luwih ala tinimbang monolith nyata — luwih alon, luwih angel di-debug, luwih akeh kanggo dioperasikake, nanging masih ora bisa dideploy kanthi independen.

Sabagéan akéh anti-pattern bali menyang wates lan basis data sing digunakake babarengan, mula njaluk bounded context lan ownership data sing bener yaiku sing mbedakake microservices sing mbantu saka sing bisa ngrusak.

Anti-pattern	Masalah
Shared database	Nggabung maneh layanan; ora bisa ngowahi skema kanthi independen
Chatty services	Akeh panggilan sync per panjaluk → latensi, kalemahan
Nano-services	Paling halus; overhead ngendang logika
No API versioning	Siji perubahan rusak saben konsumen
Synchronous chains	Latensi lan kegagalan kalilipan ing A→B→C→D
God service	Siji layanan sing kabeh gumantung (monolith anyar)