Re: Vacuum DB in Postgres Vs similar concept in other RDBMS

Поиск
Список
Период
Сортировка
От Ron Johnson
Тема Re: Vacuum DB in Postgres Vs similar concept in other RDBMS
Дата
Msg-id 4654DD66.6010407@cox.net
обсуждение исходный текст
Ответ на Re: Vacuum DB in Postgres Vs similar concept in other RDBMS  (Chris Browne <cbbrowne@acm.org>)
Ответы Re: Vacuum DB in Postgres Vs similar concept in other RDBMS  (Tom Lane <tgl@sss.pgh.pa.us>)
Re: Vacuum DB in Postgres Vs similar concept in other RDBMS  (Robert Treat <xzilla@users.sourceforge.net>)
Список pgsql-general
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 05/23/07 19:17, Chris Browne wrote:
> harpreet.dhaliwal01@gmail.com ("Harpreet Dhaliwal") writes:
>> I was just wondering if Vacuum Db in postgresql is somehow superior
>> to the ones that we have in other RDBMS.
>
> The thing that is more akin to VACUUM, in Oracle's case, is the
> rollback segment.  In Oracle, Rollback segments are areas in your
> database which are used to temporarily save the previous values when
> some updates are going on.
>
> In the case of Oracle, if a transaction rolls back, it has to go and
> do some work to clean up after the dead transaction.
>
> This is not *exactly* like PostgreSQL's notion of vacuuming, but
> that's the nearest equivalent that Oracle has.

That's the only other way to do it, no?

(Rdb/VMS has dynamically-created [made when a process attaches to
the db] Recovery Unit Journal files that store the record before-
images.)

> The Oracle InnoDB product also has the notion of rollback segments; if
> you use InnoDB tables with MySQL, the rollback functionality has much
> the same behaviour as Oracle.
>
> Note that in the case of PostgreSQL, the MVCC behaviour (which
> requires VACUUMing) has the merit that COMMIT and ROLLBACK both have
> near-zero costs; in either case, the cost is merely to mark the
> transaction as either committed or failed.  Data doesn't have to be
> touched at time of COMMIT/ROLLBACK; any costs that need to be paid are
> deferred to VACUUM time.

So it's not "near-zero cost", it's "deferred cost".

- --
Ron Johnson, Jr.
Jefferson LA  USA

Give a man a fish, and he eats for a day.
Hit him with a fish, and he goes away for good!

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGVN1mS9HxQb37XmcRAsNbAJ9hgkDpUQGVR1yxb2WrpP/m3U36eQCghv7d
9FWyD8TbSOxXiaa0e8lK5/4=
=W63C
-----END PGP SIGNATURE-----

В списке pgsql-general по дате отправления:

Предыдущее
От: "Joris Dobbelsteen"
Дата:
Сообщение: Re: Delete with subquery deleting all records
Следующее
От: Jorge Godoy
Дата:
Сообщение: Re: Geographic data sources, queries and questions