MonetDB table statistics -
below portion of statistics of 1 of tables. i'm not sure how understand width column. values in bytes? if so, know fname
, lname
have higher ascii char counts 5 , 6 , there 1 char long values in mname
.
update 1. below output of select * statistics
. i'm showing first 5 columns of ouput.
+--------+---------+------------------------+---------+-------+ | schema | table | column | type | width | +========+=========+========================+=========+=======+ | abc | targets | fname | varchar | 5 | | abc | targets | mname | varchar | 0 | | abc | targets | lname | varchar | 6 |
the column width
shows "byte-width of atom array" (defined in gdk.h
). not entire story in case of string columns, because here atom array stores offsets string heap.
monetdb uses variable-width columns, because if there few distinct string values, 64-bit offsets waste of memory. in case, fname
column needs string offsets 5 bytes, or 40 bits, , lname
needs 6 bytes (48 bits). change if new values inserted.
the 0 value mname
interesting, because width initialised 1 new columns. version using?
Comments
Post a Comment