To spread load, we will ___ the database across several nodes.
To shard a database means to horizontally partition its data across multiple servers so each holds a subset. It is the established scaling collocation: "shard by customer ID". "Slice", "split off", and "chop" are not the technical term, even though sharding is a kind of splitting. Shard works as both verb and noun (a shard) and signals horizontal scaling for write and storage capacity.
2 / 5
We ___ the primary to two read replicas for redundancy.
To replicate a database means to maintain copies of its data on other servers, usually for read scaling and high availability. The standard phrase is "replicate to read replicas". "Copy out", "duplicate over", and "clone up" are not idiomatic. Replicate pairs with replicas, lag, and primary/secondary, and is central to discussing availability and read scaling.
3 / 5
We can ___ the large table by date to keep scans fast.
To partition a table means to divide it into smaller physical pieces (for example by date range) so queries scan less data. It is the precise term: "partition by month". "Divide up", "segment off", and "cut" are vaguer. Partition applies within a single database, unlike sharding across machines, and pairs with partition keys and partition pruning.
4 / 5
Adding replicas lets us ___ without touching the primary.
To scale reads means to increase read capacity, typically by directing queries to replicas. It is the natural collocation: "scale reads with replicas", "scale writes with sharding". "Grow", "lift", and "raise reads" are not idiomatic. Pairing scale with reads, writes, or traffic is how engineers describe capacity strategy precisely.
5 / 5
If the primary dies, the cluster will ___ to a standby.
To fail over means to automatically switch to a standby system when the active one fails. It is the standard reliability collocation: "the cluster failed over in seconds" and the noun "failover". Note that "fall over" means simply to crash, which is different. "Crash over" and "drop over" are not terms. Mastering fail over versus failover (the noun) is key to discussing high availability.