Обсуждение: Re: [HACKERS] Terrible performance on wide selects

Поиск
Список
Период
Сортировка

Re: [HACKERS] Terrible performance on wide selects

От
"Dann Corbit"
Дата:
[snip]
> So (for instance) if you have 12 variable fields, you would
> store 12 integers at the start of the record.

Additionally, you could implicitly size the integers from the properties
of the column.  A varchar(255) would only need an unsigned char to store
the offset, but a varchar(80000) would require an unsigned int.

Re: [HACKERS] Terrible performance on wide selects

От
Hannu Krosing
Дата:
Dann Corbit kirjutas N, 23.01.2003 kell 02:22:
> [snip]
> > So (for instance) if you have 12 variable fields, you would
> > store 12 integers at the start of the record.
>
> Additionally, you could implicitly size the integers from the properties
> of the column.  A varchar(255) would only need an unsigned char to store
> the offset, but a varchar(80000) would require an unsigned int.

I guess that the pointer could always be 16-bit, as the offset inside a
tuple will never be more (other issues constrain max page size to 32K)

varchar(80000) will use TOAST (another file) anyway, but this will be
hidden inside the field storage in the page)

> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
--
Hannu Krosing <hannu@tm.ee>