The MonetDB columns used to actually have two "sub-columns", the head and the tail. That's why they're called "binary association tables" (BATs). These days, however, there are no more heads, or rather, the heads are only trivially increasing ranges of integral values, which do not need to actually be stored. Starting with the December 2016 release (numeric version 11.25), the storage format no longer supports storing heads, only tails.

This is where the file extensions come in:

123.head <- a "head" subcolumn for a BAT; you can no longer find these 123.tail <- a "tail" subcolumn for a BAT; _every_ column has this now

there is also the case of variable-length data, like strings or binary blobs. For these, the tail (sub)column actually stores positions within a "heap". That's why you can see

123.theap

files here and there, as well as imprints, statistical data regarding a string column for speeding up its processing:

123.timprints

The tail files are always just a plain contiguous sequence of fixed-size data, with no meta-data, gaps, etc. The (t)heap files have a more complex structure; and the structure of the imprints file I have no idea about.

Eyal

On 03/30/2017 01:53 PM, Senthil Vidhiyakar wrote:

I want to know the file system or how the data is stored in MonetDB. I know data for each column is stored in separate files. But I want to know what those .*thead *and .*tail *files inside BAT folder of the monetDB data folder mean and how data is stored in those files. Please help me ASAP.

Thank you.

---

developers-list mailing list developers-list@monetdb.org https://www.monetdb.org/mailman/listinfo/developers-list