I'm curious about what happend after execute "create index xxx" on a table of millions of rows.
According to this question Is it better to create an index before filling a table with data, or after the data is in place?,
1 million rows cost 4 seconds to create indexes, it seems fast enough.
I'm not sure if all data read and all the b+ tree index built during the 4 seconds, or the indexes are not built for whole table immediately?