Re: Lies, Damned Lies, Statistics, Marketing Departments

Read the source of the FDR (full disclosure report) for the SQL Server benchmarks. While there are customers that *might* implement a similar database design, the code underlying the test is not something that could ever be deployed in a real world environment. As implemented, you could lose power and have thousands of rows not be posted (much less committed) to the database due to a buffer implemented in the ISAPI DLL.

As to the theory that the benchmark was pulled due to "changing a single line of code" - were this the case, then the TPC would never have allowed the results to be posted. The FDR clearly indicates that the tests were based on a beta version of SQL Server. It is common industry knowledge that Microsoft changes code up to the last minute. I think they'd change it as the CD goes into the box if they could...

--
Michael D. Long
http://extremedna.homestead.com


"Don Macpherson" <nearly.donmac_at_almost-erols.com> wrote in message
news:eoz75.3972$X54.43268_at_news1.southeast.rr.com...


> David,

>     Given that the results have been withdrawn hours before a holiday

> weekend here in the US, I guess we may have to wait a while before we know

> the full story.

>

>     Of course, just like Oracle creating a "new solution" to a problem

 lead


> to TPC-D being replaced by TPC-R and TPC-H, it might be that although the

> Microsoft solution is not technically compliant with the TPC-C benchmark,

 it


> is nonetheless, a potentially viable "solution" to a customer base.  (As a

> result, I would not dismiss their solution simply because it may have

 failed


> to pass muster against a certain set of laboratory requirements)

>

>     It could also be a technicality, due to the fact that the benchmark

 was


> probably produced on code that will never be publicly released (SS7.5?).

> (SS2000 is due for release this summer, but may have had lines of source

> code changed since February, not including the line that returns the

 version


> number!  I would imagine that changing a single line of code probably

> constitutes a violation.).  The real question becomes, "What performance

> will SS2000 actually deliver?"

>

> Don Macpherson

>

> "David Pomphrey - DNP" <High.Flight_at_btinternet.com> wrote in message

> news:8jlh76$bs8$1_at_plutonium.btinternet.com...

> > Two Major MS SQL Server TPC-C benchmarks are withdrawn from the TPC.

> > ###########################################################

>

>

>