CodingLevel 6
Databases in production
Easynot started
Mediumnot started
Hardnot started
Database incidentsWhy your replicas show stale data, and how write-heavy jobs break them.3 minDatabase connection pool exhaustionWhen the bouncer stops letting people into the club.3 minHot partitions and shardingWhen you buy 10 servers, but all the traffic goes to just one.3 minDatabase index bloatWhen your fast lookups become slow because of Swiss cheese.3 minDatabase index corruptionWhen your database says a user doesn't exist, but they are right there in the table.3 minLayer 4 (TCP) load balancingHow millions of packets are routed without ever reading their contents.3 minQuery of deathOne poisoned request, retried forever, knocks over every worker that touches it.3 minDatabase autovacuum falling behindWhen ghost rows eat all your disk space and slow down queries.3 minDatabase disk saturation (IOPS)When the hard drive becomes a traffic jam and takes down the API.3 minDatabase runaway migrationHow a simple "Add Column" script can take down production for hours.3 minDatabase unreachableHandling network blips without taking down the whole API.3 minFeature pipeline data qualityGarbage in, garbage out: Automatically catching bad data before it ruins your ML model.3 minLLM output contract breaksWhy begging an LLM to "only return valid JSON" will eventually crash your backend.3 minPartition maintenance gapsWhen time-based partitioning suddenly starts dropping data on New Year's Eve.3 minTCP: the reliable byte streamAn unreliable network underneath, but every byte arrives once, in order, that's the promise TCP keeps.4 minVector index stalenessWhy semantic search engines struggle to find documents that were written 5 minutes ago.3 minWhen a node goes down, its shards re-homeKeep three copies of everything; when one machine vanishes, quietly make a fresh copy somewhere healthy.4 min