Understanding Redis Ziplist: Memory Layout, Fields, and Encoding

Redis uses a special data structure called ziplist, also known as a compressed list, to store small integers or short strings efficiently in a contiguous memory block.

The ziplist header consists of four fields: zlbytes (total bytes), zltail (offset of the tail entry), zllen (number of entries), and a variable‑length entry area ending with the zlend marker (0xFF).

Each entry stores three parts: previous_entry , which records the length of the preceding node (using 1 byte for lengths ≤254 and 5 bytes with a 0xFE prefix for larger lengths); encoding , a byte (or few bytes) whose leading two bits indicate the data type (string or integer) and the remaining bits encode the length; and p , the actual payload.

The encoding scheme defines several patterns, for example 00xxxxxx for strings up to 63 bytes, 01pppppp|qqqqqqqq for strings up to 16383 bytes, and various 11xxxxxx patterns for signed integers of different sizes.

When an entry’s length grows beyond 254 bytes, previous_entry expands from 1 to 5 bytes, which can trigger a chain reaction—called a cascade update—where subsequent entries must also adjust their previous_entry fields. This effect is rare because ziplist is intended for small data.

Overall, ziplist provides a memory‑compact representation suitable for lists with few elements, small integers, or short strings, and its design illustrates Redis’s careful balance between performance and memory efficiency.