Computer-System Operation: Interrupts, I/O, and Storage Hierarchy

Written by Rohan Nandan on May 31, 2026 · 4 min read

Operating system fundamentals are clarified by examining how hardware, I/O, and storage interact at runtime. This article summarizes the core mechanisms that keep a system responsive and coordinated.

Computer-System Operation

A modern computer system is built around shared access and concurrent activity:

• One or more CPUs and device controllers connect through a common bus and share main memory.
• CPUs and I/O devices can execute concurrently and compete for memory cycles.
• Each device controller is responsible for a specific device type and maintains a local buffer.
• Each controller type has an operating system device driver to manage it.
• The CPU moves data between main memory and controller buffers.
• I/O flows from the device to the controller buffer, then to main memory.

This layout explains why the OS is coordination-heavy: the system executes multiple activities concurrently.

Interrupts and Traps

Interrupts are events raised by hardware or software. They signal the processor to finish the current instruction and immediately handle an Interrupt Service Routine (ISR).

Key points:

• An interrupt transfers control to an ISR through the interrupt vector, which stores the addresses of service routines.
• The interrupt architecture must save the address of the interrupted instruction.
• A trap (or exception) is a software-generated interrupt caused by an error or a user request.
• Operating systems are interrupt driven: the OS reacts to events rather than polling constantly.

Bootstrapping and Startup

When a machine powers on or reboots, a bootstrap program runs first:

• It is stored in ROM or EPROM (firmware).
• It initializes system components.
• It loads the operating system kernel and starts execution.

Without a reliable bootstrap sequence, nothing else in the system can run.

I/O Control Flow

Two core I/O control styles appear in OS design:

Synchronous I/O (Blocking)

• After I/O starts, control returns to the user program only after I/O completion.
• A wait instruction can idle the CPU until the next interrupt.
• With a simple wait loop, there is contention for memory access.
• At most one I/O request is outstanding at a time (no simultaneous I/O).

Asynchronous I/O (Non-blocking)

• After I/O starts, control returns to the user program without waiting for completion.
• A system call allows the user program to wait for I/O completion if needed.
• A device-status table stores the type, address, and state of each I/O device.
• The OS indexes into this table on interrupts to update device state and completion status.

Main Memory and Secondary Storage

Main memory is the only large storage the CPU can access directly:

• Random access, typically volatile.
• Usually DRAM (Dynamic Random-Access Memory).

Secondary storage extends main memory capacity:

• Nonvolatile and large-capacity.
• Includes Hard Disk Drives (HDDs) and Solid State Drives (SSDs).
• Non-volatile memory (NVM) devices are faster than disks and increasingly common as prices drop.

Bits, Bytes, and Words

Storage is built from bits:

• A bit stores 0 or 1.
• A byte is 8 bits and is the smallest convenient unit on most systems.
• A word is the native data size of the architecture, often 64 bits (8 bytes) on modern systems.

CPUs execute most operations in word-sized chunks rather than bit by bit.

Storage Units (Binary Prefixes)

Storage is measured in bytes and their binary multiples:

• 1 KB = 1,024 bytes
• 1 MB = 1,024^2 bytes
• 1 GB = 1,024^3 bytes
• 1 TB = 1,024^4 bytes
• 1 PB = 1,024^5 bytes

Manufacturers often round these to powers of ten, but systems typically use the binary values. Networking is the exception, measuring throughput in bits rather than bytes.

Storage Hierarchy and Caching

Storage systems are organized in a hierarchy defined by:

• Speed
• Cost
• Volatility

Caching copies data into faster storage layers. Main memory can be viewed as a cache for secondary storage.

Device Drivers and OS Interfaces

Each controller type has a device driver that:

• Manages the device and its buffer.
• Provides a uniform interface between controller and kernel.

This abstraction allows the OS to treat diverse hardware consistently.

Direct Memory Access (DMA)

For high-speed devices, DMA allows the controller to transfer blocks of data directly to main memory without CPU intervention.

• Only one interrupt is generated per block, not one per byte.
• This reduces overhead and improves throughput.

Conclusion

Computer-system operation is defined by concurrency, interrupts, and layered storage. The OS orchestrates device controllers, memory, and I/O so that user programs can run smoothly and safely. These mechanisms are the foundation for everything from desktop responsiveness to high-performance server throughput.