Tag: design-database
All the articles with the tag "design-database".
-
System Design - How the Linux Page Cache Works
How the Linux page cache works: the kernel's transparent caching layer that sits between applications and disk. Covers address_space, radix tree (xarray), readahead, dirty page writeback, memory pressure eviction, and source code walkthrough from the Linux kernel.
-
System Design - How Prometheus TSDB (Time-Series Database) Works
How Prometheus stores time-series data: the in-memory head block, Write-Ahead Log, chunk encoding, compaction into persistent blocks, and how queries traverse the layered architecture. Source code walkthrough from the prometheus/prometheus repository.
-
System Design - How Vectorized Query Execution Works
How vectorized query execution works in modern databases: the Volcano model's limitations, batch-at-a-time processing, columnar memory layout, SIMD exploitation, and source code walkthrough from DuckDB and Apache DataFusion.
-
System Design - How Columnar Storage Works
How columnar storage works: row vs column layout, compression techniques, vectorized execution, Apache Parquet internals, and why modern analytics engines like ClickHouse and DuckDB are so fast.
-
System Design - How Database Replication Works
How database replication works: physical vs logical replication, MySQL binlog and PostgreSQL WAL streaming, replication lag, semi-synchronous commit, and the CAP trade-offs in real systems.
-
System Design - How Database Sharding Works
How database sharding works: partitioning strategies (hash, range, directory), routing mechanisms, resharding challenges, and how real systems (Vitess, TiDB, DynamoDB, CockroachDB) implement it.