ddia-systems

Match the data model to your access patterns: relational excels at many-to-many relationships and ad-hoc queries, document models fit self-contained aggregates, and graph models suit relationship traversal. Evaluate read/write ratio, consistency needs, and scaling path rather than popularity.

LSM trees buffer writes in memory and flush sorted files sequentially, giving high write throughput but read amplification. B-trees update pages in place, giving predictable read latency but write amplification from page splits. Use LSM for write-heavy workloads and B-trees for mixed OLTP.

Asynchronous replication lag means followers have not yet received your write when you read from them. Fix it with read-your-writes guarantees: read from the leader for recently modified data or track the replication position of your last write.

Hash partitioning distributes load evenly but destroys sort order, making range queries require scatter-gather. Key-range partitioning supports efficient range scans but risks hotspots on sequential keys like timestamps; use composite keys such as (sensor_id, date) to avoid hotspots.

Usually not. Most databases default to read committed or snapshot isolation, which permits anomalies like write skew where two transactions read the same data and write different records. Check your database's actual default and use explicit locking such as SELECT FOR UPDATE where needed.

Two-phase commit is slow, fragile, and makes the coordinator a single point of failure. Design around single-partition operations instead, and use sagas with compensating actions for cross-service operations like charging a card and reserving inventory.