DMA and Disk Hardware Features

What problem does DMA (Direct Memory Access) solve?

Reduces CPU overhead by handling I/O transfers without constant CPU intervention, only interrupting when complete.

How does DMA work?

CPU provides: operation type, memory address, device address, and data length. DMA controller manages the transfer, incrementing addresses and decrementing count until completion.

Why do modern disk drives include large RAM buffers (e.g., 8MB)?

To cache entire tracks (track buffers) and exploit spatial locality, reducing rotational latency when reading sequential sectors.

What is sector sparing/sector relocation?

Using spare sectors to replace failing sectors by copying data and remapping sector numbers, handled either by drive hardware or OS.

What is S.M.A.R.T. technology?

Self-Monitoring, Analysis and Reporting Technology that predicts disk failures by monitoring degradation of drive attributes (60% of failures are predictable mechanical issues).

What are the two classes of disk failures?

1. Unpredictable (sudden electronic/mechanical failures)

2. Predictable (gradual degradation, like head flying height)

Why are seek time and rotational latency becoming dominant performance bottlenecks?

While CPU speed and transfer rates improved exponentially (100%/year), seek/latency only improved ~10x in 20 years, now limited by physics (speed of sound at outer tracks).

What future technologies might replace disk drives?

• Pure electronic memories

• Microelectromechanical systems (MEMS)

• Hybrid Hard Drives (HHDs with flash memory)

• Windows Vista already uses flash as cache extension

How does track buffering optimize sequential reads?

Reads entire track when accessing one sector, anticipating future requests for nearby sectors to avoid full rotations.

What historical precedent suggests disks will be replaced?

Tape drives were replaced by disks as main secondary storage in 1970s; disks may similarly become tertiary storage.