Re: MVCC implementation
Date: Sat, 22 Jul 2006 13:57:39 GMT
Message-ID: <n9qwg.2479$qZ2.1109_at_trndny01>
"Marshall" <marshall.spight_at_gmail.com> wrote in message
news:1153495655.953684.44730_at_b28g2000cwb.googlegroups.com...
> Hi all,
>
> Am I correct in regarding MVCC as pretty much the state of the art
> for relational implementation? (I know we hear things about TRM
> but until there's more published it doesn't seem evaluable.)
>
> Are there any difficulties associated with MVCC implementations
> supporting nested transactions?
>
> Any good references for reading about MVCC implementation?
> It's not clear to me how it works. Googling for it is doesn't get
> me any good references; mostly manuals for SQL products
> that support it. I'm interested in learning about how to code
> one up.
>
>
Marshall,
Here I go again. I'm starting to sound like a broken record.
Take a look at this wikipedia article.
http://en.wikipedia.org/wiki/Multiversion_concurrency_control
It traces a history of MVCC, Starkey, DEC, Datatrieve, and Rdb/VMS.
I was working at DEC at the same time as Starkey, although I only met him
once, at a trade show, years later.
But I did benefit from Rdb/ELN, even though I was intending on using
Rdb/VMS.
Basically an internal version of "RDB/Eln for VMS developers" circulated around the DEC internal e-net before the Rdb/VMS product was ready for internal testing.
So I used Rdb/Eln, making sure that the port to Rdb/VMS would remain trivial. That part was easy, thanks to DEC's DSRI.
Starkey is one of the unrecognized giants, in my view. Some day, I've got to learn Firebird.
I hope you don't mind that I keep leading you back to the same place. It's not that I don't think important thinks weren't happening elsewhere at the time. It's just that this is the part of the history that I remember, and the issues you raise are almost precisely the same ones as the ones that Rdb/VMS and Starkey were wrestling with. Received on Sat Jul 22 2006 - 15:57:39 CEST