Reference · 19 terms
Glossary
Every short form used across the series. Inside a lesson, hover or tap any dotted term to see the same entry in place.
Case-study ideas
- Bloom filterBloom filter
- A small structure that says 'definitely not seen' or 'probably seen'. It never forgets an item but can give false positives.Lesson 0 →
- cache hit ratioCache hit ratio
- The share of requests answered from the cache. Higher means less load on the origin.Lesson 4 →
- chunkingChunking
- Cutting a file into pieces so each piece can be stored, hashed and synced on its own.Lesson 2 →
- consistent hashingConsistent hashing
- Place nodes and keys on a ring so that adding or removing a node moves only a small share of the keys.Lesson 1 →
- count-min sketchCount-min sketch
- A small table of counters that gives approximate counts for a huge stream, never counting too low.Lesson 5 →
- CRDTCRDT
- A data type designed so that replicas that apply the same updates in any order end up in the same state.Lesson 3 →
- deduplicationDeduplication
- Storing identical data once and pointing to it from many places.Lesson 2 →
- idempotent ingestionIdempotent ingestion
- Running the same import twice gives the same result, so retries are safe.Lesson 10 →
- operational transformationOperational transformation (OT)
- Adjust an edit to account for edits that happened at the same time, so everyone converges on the same text.Lesson 3 →
- order bookOrder book
- The lists of buy and sell orders for a stock, grouped by price.Lesson 8 →
- politenessPoliteness
- Limits that stop a crawler from overloading one website: robots.txt rules and a delay between requests to the same host.Lesson 0 →
- price-time priorityPrice-time priority
- The best price matches first; at the same price, the earlier order matches first.Lesson 8 →
- quorumQuorum
- Read from R and write to W of N replicas with R + W > N, so every read overlaps the latest write.Lesson 1 →
- RPORPO (recovery point objective)
- How much recent data you can afford to lose in a failure, measured in time.Lesson 7 →
- RTORTO (recovery time objective)
- How long you can afford to be down before service is restored.Lesson 7 →
- Snowflake IDSnowflake ID
- A 64-bit ID made of a timestamp, a machine ID and a sequence number, so IDs are unique and roughly time-ordered.Lesson 6 →
- split brainSplit brain
- Two parts of a system both think they are in charge after a network split, and accept conflicting writes.Lesson 7 →
- URL frontierURL frontier
- The queue of URLs a crawler still has to visit, ordered by priority and by politeness limits per host.Lesson 0 →
- version vectorVersion vector
- A small map of counters that lets replicas tell whether one update came after another or the two were concurrent.Lesson 1 →