The number of primary shards is fixed when the index is created and cannot be changed. The default of five was chosen for a different era of cluster sizes and is wrong for most indices in both directions.
PUT /products
{ "settings": { "number_of_shards": 1, "number_of_replicas": 1 } }Each shard is a Lucene index with its own memory, file handles and merge overhead, so a 200 MB index split five ways is five times the overhead for no benefit — and a query fans out to all five and waits for the slowest. The rule of thumb is a shard of 10-50 GB, which for most application indices means one. Getting it wrong is survivable because reindexing behind an alias is routine; not having the alias is what makes it painful.