Also pretty sure almost every database with variable width columns has this issue. An old MS SQL Server optimization was to use chars and not have nulls since t

Also pretty sure almost every database with variable width columns has this issue. An old MS SQL Server optimization was to use chars and not have nulls since that drops a few bits too. That way your table was easy to find row 500 on disk. This was back when SCSI and IDE drives were common though. Doesn't indexes fix this too? You can index on column 100 which puts in on disk as Column100|PrimaryKey and now it's a really narrow table. Like yeah you shouldn't have a table with 1,500 columns but sometimes you really do need it. Edit Flipped the index layout.