Howard J. Rogers wrote:
> "Daniel Morgan" <damorgan_at_x.washington.edu> wrote in message
> news:1079267535.385345_at_yasure...
>
>>Howard J. Rogers wrote:
>>
>>
>>>Well, let me just ask you this then. Given that OPS first made an
>>>(admittedly rather faltering) appearance in Oracle 6, and that there was
>>>therefore the best part of 10 years' development effort to get OPS to
>
> where
>
>>>it was in 8.1.7, do you honestly believe that they trashed the entire
>
> lot
>
>>>and started with an utterly clean slate in order to get RAC out of the
>
> door?
>
>>On the clustering part no? On the cache fusion part yes.
>
>
> Ah. I think that is a bit of a, shall I say, "clarification" on your part.
>
>
>>>Particularly when you might care to bear in mind that cache fusion
>
> actually
>
>>>made an appearance in 8.1.7 regarding consistent read buffer transfers.
>>>
>>>But as a mere humble trainer, it was nevertheless strictly verboten to
>>>discuss RAC as a developement to OPS. I imagine the particular
>
> developers
>
>>>you plied with mind-altering liquids were under similar injunctions from
>
> the
>
>>>top.
>>
>>Given other parts of the, how can I say this gently, 'wide ranging'
>>conversation, I don't think it would have mattered what they had been
>>told to say and not say. And there was a lot of evidence of, how can
>>I say this gently, 'a lack of inhibition'.
>>
>>In other words ... yeah I believed them. And that is not my nature.
>>
>>>I realise it won't be worth anything to you, but I'll give you a
>
> categorical
>
>>>assurance that the codebase for RAC is the same as the one for OPS, with
>>>enhancements of course.
>>>
>>>HJR
>>
>>Enhancements don't turn block pinging into block sharing. That's quite a
>>bit more than an enhancement. Have you seen the huge number of patents
>>Oracle filed on RAC to keep others from duplicating it?
>
>
> But as I said, block transfer across the interconnect wasn't even new in 9i,
> because it was there in 8.1.7 for consistent read transfers.
>
> Or did you miss that bit?
>
> All they did in 9i compared to 8.1.7 was to make block transfer across the
> interconnect the (default) mechanism for all block transfers rather than
> just some.
>
> Do I believe that at some point new code had to be cut to make any form of
> interconnect-based block transfer possible? Obviously. But it wasn't brand
> new in 9i, because the principle had been established before hand.
>
> HJR
The stuff in you say in 8i was an attempt to fix the problem before
they threw in the towel.
--
Daniel Morgan
http://www.outreach.washington.edu/ext/certificates/oad/oad_crs.asp
http://www.outreach.washington.edu/ext/certificates/aoa/aoa_crs.asp
damorgan_at_x.washington.edu
(replace 'x' with a 'u' to reply)
Received on Sun Mar 14 2004 - 13:22:36 CST
