Kepiye panggantosan page mlaku, lan apa sing nyebabake thrashing?

Question

IT Interview Practice · Accepted Answer

Nalika RAM fisik kebak nanging proses butuh page maneh, OS kudu **nundhung (evict)** page sing manggon kanggo nggawe papan — nulis dheweke menyang **swap** (disk) yen kotor (dirty). Page endi sing ditundhung iku keputusan **page replacement**, lan yen salah bakal ngrusak performa.

## Kabijakan panggantosan

Optimum teoritis yaiku nundhung page sing bakal dienggo paling adoh ing mangsa ngarep (**OPT-e Bélády**) — mokal, amarga kowe ora bisa ndeleng mangsa ngarep. Mula kita nyedhaki **LRU (least recently used)**: nundhung page sing paling suwe ora dienggo. LRU sing tenanan butuh timestamp ing saben akses (kelarangen), mula kernel nyata nganggo **algoritma clock (second-chance)**:

```text
Frames arranged in a ring; each has a REFERENCE bit the hardware sets on access.
A "hand" sweeps around the ring:
   ref bit = 1 → clear it, give the page a second chance, move on
   ref bit = 0 → evict THIS page (approximates "least recently used")

[1]→[1]→[0] ← evict here   (the sweep clears 1s until it finds a 0)
         ▲___________|
```

Clock menehi asil kaya LRU kanthi biaya O(1). Linux nyempurnakake maneh nganggo **daftar LRU active/inactive**.

## Working set lan thrashing

**Working set** sawijining proses iku sakumpulan page sing dienggo kanthi aktif ing sajroning jendhela wektu. Paging mlaku apik *anggere gunggung working set pas ing RAM*. Nalika ora pas:

```text
working set > RAM → evict a page that's about to be needed → immediate page fault →
evict another needed page → ... → the CPU spends ~all its time paging to disk, ~0% on real work
                            = THRASHING
```

Thrashing iku jurang, dudu ereng-erengan: ngliwati wates sithik wae lan throughput ambruk amarga meh saben akses miss. Gejala: CPU meh nganggur nanging I/O disk jenuh, lan latency mbledhos.

## Ndandani / ngindhari

- **Tambah RAM** utawa **kurangi beban** (proses bebarengan luwih sithik) supaya working set pas.
- Scheduler bisa **nyuspend lan nge-swap out** kabeh proses kanggo ngurangi panjaluk (medium-term scheduling).
- **OOM killer** Linux iku pilihan pungkasan — mateni proses tinimbang thrash salawase.
- Nambah **locality** supaya working set-e dhewe luwih cilik.

## Kenapa iki penting

Iki nggandheng kabeh crita memori dadi siji lan dadi pitakonan senior sing disenengi amarga thrashing iku mode gagal produksi sing nyata: service sing apik ing N pangganggo tiba saka jurang ing N+1 nalika working set-e ngluwihi RAM. Ngenali tandhane (paging abot, disk jenuh, CPU mandheg) lan ngerti tuas-tuase (RAM, ngurangi beban, locality — lan yen swap ngowahi kekurangan memori dadi bencana latency) iku bedane antara ndiagnosa ing sawetara menit lan mung ndeleng kothak sing beku kanthi misterius.

Kabijakan panggantosan

Working set lan thrashing

Ndandani / ngindhari

Kenapa iki penting

Pitakon sing gegandhengan