Yaya maye gurbin page ke aiki, kuma me ke haifar da thrashing?

Question

IT Interview Practice · Accepted Answer

Idan RAM ta zahiri ta cika amma processes suna bukatar ƙarin pages, OS dole ya **kori (evict)** page mazauni don samar da sarari — yana rubuta shi zuwa **swap** (disk) idan yana da datti (dirty). Wanne page za a kori shi ne shawarar **page replacement**, kuma yin sa ba daidai ba yana lalata aiki (performance).

## Manufofin maye gurbi

Mafi kyawun ka'idar shine kori page da za a yi amfani da shi mafi nisa a nan gaba (**OPT na Bélády**) — ba zai yiwu ba, tunda ba za ka iya ganin gaba ba. Don haka muna kusanto **LRU (least recently used)**: kori page da ba a yi amfani da shi mafi dadewa ba. LRU na gaske yana bukatar timestamp a kowane damar shiga (mai tsada sosai), don haka kernels na gaske suna amfani da **clock (second-chance) algorithm**:

```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 yana ba da sakamako kamar LRU a farashin O(1). Linux yana ƙara gyara shi da **active/inactive LRU lists**.

## Working set da thrashing

**Working set** na process shi ne saitin pages da yake amfani da su sosai a cikin taga lokaci. Paging yana aiki lafiya *muddin haɗaɗɗun working sets sun dace a RAM*. Idan ba su dace ba:

```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 dutse ne mai tsauri, ba gangare ba: ka wuce iyaka kaɗan sai throughput ya rushe domin kusan kowane damar shiga yana miss. Alamomi: CPU kusan babu aiki amma disk I/O cike, kuma latency yana fashewa.

## Gyara / kaucewa masa

- **Ƙara RAM** ko **rage load** (ƙarancin processes lokaci guda) don working sets su dace.
- Scheduler zai iya **dakatarwa da swap out** ɗin cikakkun processes don rage buƙata (medium-term scheduling).
- **OOM killer** na Linux shi ne mafita ta ƙarshe — kashe process maimakon thrash har abada.
- Inganta **locality** don working set ɗin kansa ya zama ƙarami.

## Me ya sa yake da muhimmanci

Wannan yana haɗa dukkan labarin memory tare kuma tambaya ce da ake so a matakin senior domin thrashing yanayin gazawa ne na gaske a production: sabis da yake lafiya a N masu amfani yana faɗuwa daga dutse a N+1 yayin da working set ɗinsa ya wuce RAM. Gane sa hannun (paging mai nauyi, disk cike, CPU tsaye) da sanin maɓallan (RAM, load shedding, locality — da cewa swap yana mai da ƙarancin memory zuwa masifar latency) shi ne bambanci tsakanin gano shi cikin mintuna da zama kana kallon akwatin da ya daskare cikin mamaki.

Manufofin maye gurbi

Working set da thrashing

Gyara / kaucewa masa

Me ya sa yake da muhimmanci

Tambayoyi masu alaƙa