Counting Web Page Visits with Redis: Hash, Bitset, and HyperLogLog Techniques

Pinduoduo offers high salaries and intense work, and its interview includes a question: how to use Redis to count website user visits?

Using Hash

Hash is a basic Redis data structure that maps keys to a hash table; on collisions it creates a linked list.

When a user visits, use the user ID if logged in, otherwise generate a random key to identify the user. Use the HSET command with a key composed of the URI and date, a field of the user ID or random identifier, and a value of 1.

To obtain the daily visit count for a page, use HLEN on the hash.

Pros: Simple, easy to implement, fast queries, high data accuracy.

Cons: High memory consumption; performance degrades as keys increase, making it unsuitable for billions of page views.

Using Bitset

A 32‑bit integer can represent 32 users when each bit denotes a user. Converting IDs to bits saves 32× memory. For large datasets, a bitset can drastically reduce memory usage; for 100 million users it requires about 12 MB.

Redis provides SETBIT to mark a user’s visit, GETBIT to check if a user visited, and BITCOUNT to count daily visits.

Pros: Very low memory usage, fast queries, can query specific users.

Cons: Slight data inaccuracies for anonymous users; requires mapping non‑logged‑in users to IDs, adding overhead.

Using Probabilistic Algorithm

When exact counts are unnecessary, HyperLogLog provides an approximate cardinality with minimal memory (about 12 KB per key) and a typical error of ~0.81%.

Use PFADD on each visit and later retrieve the estimate with PFCOUNT.

Pros: Extremely low memory, suitable for massive traffic sites.

Cons: Cannot query individual users; results have a small error margin.

These three methods—Hash, Bitset, and HyperLogLog—cover common approaches to counting website visits with Redis.