CS-3310 - 03 Memory, 04 I/O and Disks - Extra Concepts

Question

Answer

What is segmentation in memory management?

A scheme dividing memory into variable-sized logical segments.

How does segmentation differ from paging?

Segments vary in size; pages are fixed size.

Why combine segmentation and paging?

Segmentation gives logical organization, paging provides efficiency and protection.

What is a segment fault?

An illegal access outside a segment's limit.

What problem does the TLB solve?

It caches recent address translations for faster lookup.

What is a TLB shootdown?

Invalidation of TLB entries across multiple CPUs after a mapping change.

What is demand paging?

Loading a page into memory only when it is first accessed.

What is prepaging?

Loading additional pages the OS predicts will be used soon.

Describe the steps in a page-fault handling sequence.

Trap, locate page on disk, allocate frame, read page, update tables, resume.

What is the working set model?

A set of pages a process actively uses; helps reduce thrashing.

How does WSClock improve on simple LRU?

It combines clock replacement with working-set tracking.

What is copy-on-write (COW)?

A mechanism that delays copying shared pages until a write occurs.

What is memory compression used for?

Reducing swap I/O by storing compressed pages in RAM.

What is NUMA and why does it matter?

Non-Uniform Memory Access; access time depends on which CPU node owns the memory.

What is page coloring?

A cache-aware page allocation strategy to minimize cache conflicts.

How does shared memory differ from private memory?

Shared memory regions are accessible by multiple processes.

What are the main stages of an I/O operation?

Request, scheduling, execution, completion.

What is an I/O request queue?

A per-device queue storing pending read/write requests.

What is interrupt coalescing?

Combining multiple interrupts into one to reduce overhead.

What is asynchronous I/O?

I/O where processes continue running while operations complete.

What is the benefit of direct I/O?

Avoids OS buffering for faster transfers.

What is zero-copy I/O?

Transfers data directly between disk and user space without copies.

What is an upper-half and lower-half driver?

Upper handles high-level logic, lower interacts with hardware.

What is hot-plugging?

Adding or removing hardware while the system is running.

What is PnP (Plug and Play)?

Automatic hardware detection and configuration by the OS.

What is hybrid polling?

A mix of polling and interrupts for high-speed devices.

Why might an OS reorder I/O requests?

To minimize seek time and latency.

What is I/O virtualization?

Allowing virtual machines to share or emulate I/O devices.

What key difference distinguishes SSDs from HDDs?

SSDs have no moving parts.

What is the Flash Translation Layer (FTL)?

Firmware mapping logical to physical flash addresses.

What is wear leveling?

Technique to evenly distribute writes on flash memory.

What is the TRIM command used for?

Informing SSD which blocks are unused for reuse.

What is NCQ (Native Command Queuing)?

Feature allowing disks to handle multiple queued requests efficiently.

What are IOPS?

Input/Output operations per second.

How do SSDs differ in failure mode from HDDs?

SSDs lose entire blocks; HDDs lose sectors gradually.

What is disk caching?

Using onboard RAM in drives to buffer data.

What is SMR (Shingled Magnetic Recording)?

A method to overlap tracks for higher density.

What is MTBF?

Mean time between failures; reliability measure.

What is RAID-Z?

ZFS's improved parity RAID for fault tolerance.

What is advanced format?

Using 4KB physical sectors instead of 512B.

What is the purpose of the buffer cache?

To store frequently accessed file data in memory.

What is write-back caching?

Delaying disk writes until later to improve performance.

What is write-through caching?

Writing data immediately to ensure consistency.

What are access control lists (ACLs)?

Extended permissions allowing fine-grained access control.

What does the 'noexec' mount option do?

Prevents execution of binaries from the filesystem.

What is the dentry cache used for?

To speed up file name lookups.

What is a log-structured file system?

A file system that writes data sequentially like a log.

What is delayed allocation?

Deferring block assignment to optimize layout.

What is defragmentation?

Rearranging data to make files contiguous.

What is an atime update?

A record of the last access time for a file.

What is an extended attribute (xattr)?

Extra metadata stored alongside a file.

What is the benefit of file checksums?

Detecting and repairing data corruption.

What is copy-on-write in file systems?

Writing new blocks instead of overwriting old ones.

What is journaling metadata-only mode?

Only filesystem metadata is logged for recovery.

What is soft updates?

Ordered writes to maintain consistency without journaling.

What is a bind mount?

Re-mounting a directory at another path.