File Access Methods

Files are stored as bytes; access methods are the patterns programs use to read them.

1. Sequential Access

Read or write bytes in order. Operations: read_next, write_next, reset. Matches tape model. Most logs and streams use this.

2. Direct (Random) Access

Treat the file as a numbered sequence of blocks; jump to block n in O(1). Required for databases and large files. lseek(fd, offset, SEEK_SET) is the system call.

3. Indexed Access

Maintain an index that maps a key (e.g., student ID) to a block. Looking up a record means searching the index then jumping to the block. Used in ISAM and many DBMS file formats.

Directory Structures

Linux uses a tree with hard links and symbolic links (effectively producing an acyclic graph for files).

Mounting

A filesystem must be mounted at a directory (the mount point) before its files are accessible. mount /dev/sdb1 /mnt/usb overlays the root of the new filesystem onto /mnt/usb.

Disk Structure

A disk is divided into:

• Boot block — bootloader code.
• Superblock — filesystem metadata (size, free counts).
• Inode table — fixed-size array of inodes.
• Data blocks — file contents.
• Free-list / bitmap — tracks free blocks.

Allocation Methods

• Contiguous — file occupies consecutive blocks. Fast random access; suffers external fragmentation.
• Linked — each block points to the next. No fragmentation; poor random access.
• Indexed — an index block holds pointers to all data blocks. Combines random access with no external fragmentation. Used in Unix-like inodes (with multi-level indirection).

File Protection — General

• Access lists (ACLs) — per-user permission entries.
• Group / role-based — Unix's owner/group/other model.
• Capabilities — unforgeable tokens, used in microkernels.

Summary

• Sequential, direct, and indexed access methods cover most workloads.
• Modern systems use a tree directory with hard/symbolic links.
• Indexed allocation (Unix inode) is the practical sweet spot for disk layout.