Re: computational model of transactions
Date: Fri, 04 Aug 2006 20:46:41 GMT
Message-ID: <RmOAg.1142$1f6.384_at_newssvr27.news.prodigy.net>
No. MVCC ensures the consistency of a sequence of statements that only read from the database by hiding the results of any writes by any other transactions outstanding during the read. By enclosing reads in a MVCC transaction, it appears that the entire set of reads occurs instantaneously, drawing only from the state of the database that was in effect when the transaction started.
"Marshall" <marshall.spight_at_gmail.com> wrote in message
news:1154712384.843649.326550_at_b28g2000cwb.googlegroups.com...
> Erwin wrote:
>>
>> AFAICT, MVCC is only safe to use in transactions that don't update
>> anything, i.e. transactions that do *nothing but reads*.
>
> Um, if we are doing nothing but reads, we don't need any concurrency
> controls at all. Everything can read everything all the time safely, if
> there are no writes going on. So it kind of seems like above you're
> saying MVCC is a no-op.
>
>
> Marshall
>
Received on Fri Aug 04 2006 - 22:46:41 CEST