Except, of course, that if you update the row, you have to re-write the blob. When any part of the row is changed the WHOLE row is re-written. This means that if you put a 4 GB blob as the last column of a table with 4 other data fields before it, whenever you update one of the other 4 fields you will re-write the entire record INCLUDING the 4 GB blob which you will have to both read into memory and write back out to disk, perhaps multiple times. This is one of the reasons that large blobs (and text fields, which are the same thing, just with overloaded meaning) should be stored in their own table.