Re: On-Disk Tuple Size

[Moved from general to -hackers.]

On Sat, 20 Apr 2002, Martijn van Oosterhout wrote:

> > In MS SQL server, for example....
>
> Where is the information needed to determine visibility for transactions? In
> Postgres that's at least 16 bytes (cmin,cmax,xmin,xmax). How does SQL server
> do that?

SQL Server doesn't use MVCC; it uses locking. (This is not necessarially
less advanced, IMHO; it has the nice properties of saving a bunch of
space and ensuring that, when transaction isolation is serializable,
commits won't fail due to someone else doing updates. But it has costs,
too, as we all know.)

> > (If there were variable length columns, you would have after this:
> > two bytes for the number of columns, 2 bytes per column for the
> > data offsets within the tuple, and then the variable data.)
>
> In postgres, variable length columns don't cost anything if you don't use
> them.

Right; just as in SQL server. This was just sort of a side note
for those who are curious.

> An int is always 4 bytes, even if there are variable length columns
> elsewhere. The only other overhead is 4 bytes for the OID....

Which would be good to get rid of, if we can.

> ...and 6 bytes for the CTID, which I guess may be unnecessary.

Really? How would things work without it?

cjs
--
Curt Sampson <cjs(at)cynic(dot)net> +81 90 7737 2974 http://www.netbsd.org
Don't you know, in this new Dark Age, we're all light. --XTC